State-of-the-art methods for Human-AI Teaming and Zero-shot Cooperation focus on task completion i.e. task rewards, as the sole evaluation metric while being agnostic to `how' the two agents work with each other. Furthermore, subjective user studies only offer limited insight into the quality of cooperation existing within the team. Specifically, we are interested in understanding the cooperative behaviors arising within the team when trained agents are paired with humans - a problem that has been overlooked by the existing literature. To formally address this problem, we propose the concept of constructive interdependence - measuring how much agents rely on each other’s actions to achieve the shared goal - as a key metric for evaluating cooperation in human-agent teams. We measure interdependence in terms of action interactions in a STRIPS formalism, and define metrics that allow us to assess the degree of reliance between the agents' actions. We pair state-of-the-art agents with learned human models as well as human participants in a user study for the popular Overcooked domain, and evaluate the task reward and teaming performance for these human-agent teams. While prior work has claimed that state-of-the-art agents exhibit cooperative behavior based on their high task rewards, our results reveal that these agents often fail to induce cooperation, as evidenced by consistently low interdependence across teams. Furthermore, our analysis reveals that teaming performance is not necessarily correlated with task reward, highlighting that task reward alone cannot reliably measure cooperation arising in a human-agent team.

Recent advances in reasoning-oriented Large Language Models (LLMs) have been driven by the introduction of Chain-of-Thought (CoT) traces, where models generate intermediate reasoning traces before producing an answer.These traces, as in DeepSeek R1, are not only used to guide model inference but also serve as supervision signals for Knowledge Distillation (KD) to improve smaller models.A prevailing but under-examined implicit assumption is that these CoT traces when emitted at inference time are both semantically correct and interpretable for the end-users.While there are reasons to believe that these intermediate tokens help improve solution accuracy, in this work, we question their validity (semantic correctness) and interpretability to the end user.To isolate the effect of trace semantics, we design experiments in the Question Answering (QA) domain using a rule-based problem decomposition method.This enables us to create Supervised Fine-Tuning (SFT) datasets for LLMs where -each QA problem is paired with either verifiably correct or incorrect CoT traces, while always providing the correct final solution.Trace correctness at inference time is then evaluated by checking the accuracy of every sub-step in decomposed reasoning chains.To assess end-user interpretability, we finetune LLMs with three additional types of CoT traces: R1 traces, R1 trace summaries, and post-hoc explanations of R1 traces.We further conduct a human-subject study with 100 participants asking them to rate the interpretability of each trace type on a standardized Likert scale.Our experiments reveal two key findings -(1) CoT trace correctness is not reliably correlated with the model's generation of correct final answers: correct traces led to correct solutions only for 28% test-set problems while incorrect traces don't necessarily degrade solution accuracy.(2) In end-user interpretability studies, fine-tuning * Work done while a PhD student at ASU, currently at Samsung Research America on verbose R1 traces prooduced the best model performance but these traces were rated as least interpretable by users, scoring on average 3.39 for interpretability and 4.59 for cognitive load metrics on a 5-point Likert scale.In contrast, the decomposed traces that are judged significantly more interpretable don't lead to comparable solution accuracy.Together, these findings challenge the assumption in question suggesting that researchers and practitioners should decouple model supervision objectives from end-user-facing trace design. 1

There has been considerable divergence of opinion on the reasoning abilities of Large Language Models (LLMs). While the initial optimism that reasoning might emerge automatically with scale has been tempered thanks to a slew of counterexamples--ranging from multiplication to simple planning--there persists a wide spread belief that LLMs can self-critique and improve their own solutions in an iterative fashion. This belief seemingly rests on the assumption that verification of correctness should be easier than generation--a rather classical argument from computational complexity--which should be irrelevant to LLMs to the extent that what they are doing is approximate retrieval. In this paper, we set out to systematically investigate the effectiveness of iterative prompting in the context of reasoning and planning. We present a principled empirical study of the performance of GPT-4 in three domains: Game of 24, Graph Coloring, and STRIPS planning. We experiment both with the model critiquing its own answers and with an external correct reasoner verifying proposed solutions. In each case, we analyze whether the content of criticisms actually affects bottom line performance, and whether we can ablate elements of the augmented system without losing performance. We observe significant performance collapse with self-critique and significant performance gains with sound external verification. We also note that merely re-prompting with a sound verifier maintains most of the benefits of more involved setups.

One of the core challenges of explaining decisions made by modern AI systems is the need to address the potential gap in the inferential capabilities of the system generating the decision and the user trying to make sense of it. This inferential capability gap becomes even more critical when it comes to explaining sequential decisions. While there have been some isolated efforts at developing explanation methods suited for complex decision-making settings, most of these current efforts are limited in scope. In this paper, we introduce a general framework for generating explanations in the presence of inferential capability gaps. A framework that is grounded in the generation of simplified representations of the agent model through the application of a sequence of model simplifying transformations. This framework not only allows us to develop an extremely general explanation generation algorithm, but we see that many of the existing works in this direction could be seen as specific instantiations of our more general method. While the ideas presented in this paper are general enough to be applied to any decision-making framework, we will focus on instantiating the framework in the context of stochastic planning problems. As a part of this instantiation, we will also provide an exhaustive characterization of explanatory queries and an analysis of various classes of applicable transformations. We will evaluate the effectiveness of transformation-based explanations through both synthetic experiments and user studies.

In scenarios involving robots generating and executing plans, conflicts can arise between cost-effective robot execution and meeting human expectations for safe behavior. When humans supervise robots, their accountability increases, especially when robot behavior deviates from expectations. To address this, robots may choose a highly constrained plan when monitored and a more optimal one when unobserved. While this behavior is not driven by human-like motives, it stems from robots accommodating diverse supervisors. To optimize monitoring costs while ensuring safety, we model this interaction in a trust-based game-theoretic framework. However, pure-strategy Nash equilibrium often fails to exist in this model. To address this, we introduce the concept of a trust boundary within the mixed strategy space, aiding in the discovery of optimal monitoring strategies. Human studies demonstrate the necessity of optimal strategies and the benefits of our suggested approaches.

In this work, we design an Artificially Intelligent Task Allocator (AITA) that proposes a task allocation for a team of humans. A key property of this allocation is that when an agent with imperfect knowledge (about their teammate's costs and/or the team's performance metric) contests the allocation with a counterfactual, a contrastive explanation can always be provided to showcase why the proposed allocation is better than the proposed counterfactual. For this, we consider a negotiation process that produces a negotiation-aware task allocation and, when contested, leverages a negotiation tree to provide a contrastive explanation. With human subject studies, we show that the proposed allocation indeed appears fair to a majority of participants and, when not, the explanations generated are judged as convincing and easy to comprehend.

Our work aims at efficiently leveraging ambiguous demonstrations for the training of a reinforcement learning (RL) agent. An ambiguous demonstration can usually be interpreted in multiple ways, which severely hinders the RL agent from learning stably and efficiently. Since an optimal demonstration may also suffer from being ambiguous, previous works that combine RL and learning from demonstration (RLfD works) may not work well. Inspired by how humans handle such situations, we propose to use self-explanation (an agent generates explanations for itself) to recognize valuable high-level relational features as an interpretation of why a successful trajectory is successful. This way, the agent can leverage the explained important relations as guidance for its RL learning. Our main contribution is to propose the Self-Explanation for RL from Demonstrations (SERLfD) framework, which can overcome the limitations of existing RLfD works. Our experimental results show that an RLfD model can be improved by using our SERLfD framework in terms of training stability and performance. To foster further research in self-explanation-guided robot learning, we have made our demonstrations and code publicly accessible at https://github.com/YantianZha/SERLfD. For a deeper understanding of our work, interested readers can refer to our arXiv version at https://arxiv.org/pdf/2110.05286.pdf, including an accompanying appendix.

We argue that auto-regressive LLMs cannot, by themselves, do planning or self-verification (which is after all a form of reasoning), and shed some light on the reasons for misunderstandings in the literature. We will also argue that LLMs should be viewed as universal approximate knowledge sources that have much more meaningful roles to play in planning/reasoning tasks beyond simple front-end/back-end format translators. We present a vision of LLM-Modulo Frameworks that combine the strengths of LLMs with external model-based verifiers in a tighter bi-directional interaction regime. We will show how the models driving the external verifiers themselves can be acquired with the help of LLMs. We will also argue that rather than simply pipelining LLMs and symbolic components, this LLM-Modulo Framework provides a better neuro-symbolic approach that offers tighter integration between LLMs and symbolic components, and allows extending the scope of model-based planning/reasoning regimes towards more flexible knowledge, problem and preference specifications.

Large language model (LLM) performance on reasoning problems typically does not generalize out of distribution. Previous work has claimed that this can be mitigated with chain of thought prompting--a method of demonstrating solution procedures--with the intuition that it is possible to in-context teach an LLM an algorithm for solving the problem. This paper presents a case study of chain of thought on problems from Blocksworld, a classical planning domain, and examines the performance of two state-of-the-art LLMs across two axes: generality of examples given in prompt, and complexity of problems queried with each prompt. While our problems are very simple, we only find meaningful performance improvements from chain of thought prompts when those prompts are exceedingly specific to their problem class, and that those improvements quickly deteriorate as the size n of the query-specified stack grows past the size of stacks shown in the examples. We also create scalable variants of three domains commonly studied in previous CoT papers and demonstrate the existence of similar failure modes. Our results hint that, contrary to previous claims in the literature, CoT's performance improvements do not stem from the model learning general algorithmic procedures via demonstrations but depend on carefully engineering highly problem specific prompts. This spotlights drawbacks of chain of thought, especially the sharp tradeoff between possible performance gains and the amount of human labor necessary to generate examples with correct reasoning traces.

Trust between team members is an essential requirement for any successful cooperation. Thus, engendering and maintaining the fellow team members' trust becomes a central responsibility for any member trying to not only successfully participate in the task but to ensure the team achieves its goals. The problem of trust management is particularly challenging in mixed human-robot teams where the human and the robot may have different models about the task at hand and thus may have different expectations regarding the current course of action, thereby forcing the robot to focus on the costly explicable behavior. We propose a computational model for capturing and modulating trust in such iterated human-robot interaction settings, where the human adopts a supervisory role. In our model, the robot integrates human's trust and their expectations about the robot into its planning process to build and maintain trust over the interaction horizon. By establishing the required level of trust, the robot can focus on maximizing the team goal by eschewing explicit explanatory or explicable behavior without worrying about the human supervisor monitoring and intervening to stop behaviors they may not necessarily understand. We model this reasoning about trust levels as a meta reasoning process over individual planning tasks. We additionally validate our model through a human subject experiment.

Dealing with planning problems with both logical relations and numeric changes in real-world dynamic environments is challenging. Existing numeric planning systems for the problem often discretize numeric variables or impose convex constraints on numeric variables, which harms the performance when solving problems, especially when the problems contain obstacles and non-linear numeric effects. In this work, we propose a novel algorithm framework to solve numeric planning problems mixed with logical relations and numeric changes based on gradient descent. We cast the numeric planning with logical relations and numeric changes as an optimization problem. Specifically, we extend the syntax to allow parameters of action models to be either objects or real-valued numbers, which enhances the ability to model real-world numeric effects. Based on the extended modeling language, we propose a gradient-based framework to simultaneously optimize numeric parameters and compute appropriate actions to form candidate plans. The gradient-based framework is composed of an algorithmic heuristic module based on propositional operations to select actions and generate constraints for gradient descent, an algorithmic transition module to update states to the next ones, and a loss module to compute loss. We repeatedly minimize loss by updating numeric parameters and compute candidate plans until it converges into a valid plan for the planning problem.

There is a growing interest in applying pre-trained large language models (LLMs) to planning problems. However, methods that use LLMs directly as planners are currently impractical due to several factors, including limited correctness of plans, strong reliance on feedback from interactions with simulators or even the actual environment, and the inefficiency in utilizing human feedback. In this work, we introduce a novel alternative paradigm that constructs an explicit world (domain) model in planning domain definition language (PDDL) and then uses it to plan with sound domain-independent planners. To address the fact that LLMs may not generate a fully functional PDDL model initially, we employ LLMs as an interface between PDDL and sources of corrective feedback, such as PDDL validators and humans. For users who lack a background in PDDL, we show that LLMs can translate PDDL into natural language and effectively encode corrective feedback back to the underlying domain model. Our framework not only enjoys the correctness guarantee offered by the external planners but also reduces human involvement by allowing users to correct domain models at the beginning, rather than inspecting and correcting (through interactive prompting) every generated plan as in previous work. On two IPC domains and a Household domain that is more complicated than commonly used benchmarks such as ALFWorld, we demonstrate that GPT-4 can be leveraged to produce high-quality PDDL models for over 40 actions, and the corrected PDDL models are then used to successfully solve 48 challenging planning tasks. Resources, including the source code, are released at: https://guansuns.github.io/pages/llm-dm.

Generating plans of action, and reasoning about change have long been considered a core competence of intelligent agents. It is thus no surprise that evaluating the planning and reasoning capabilities of large language models (LLMs) has become a hot topic of research. Most claims about LLM planning capabilities are however based on common sense tasks–where it becomes hard to tell whether LLMs are planning or merely retrieving from their vast world knowledge. There is a strong need for systematic and extensible planning benchmarks with sufficient diversity to evaluate whether LLMs have innate planning capabilities. Motivated by this, we propose PlanBench, an extensible benchmark suite based on the kinds of domains used in the automated planning community, especially in the International Planning Competition, to test the capabilities of LLMs in planning or reasoning about actions and change. PlanBench provides sufficient diversity in both the task domains and the specific planning capabilities. Our studies also show that on many critical capabilities–including plan generation–LLM performance falls quite short, even with the SOTA models. PlanBench can thus function as a useful marker of progress of LLMs in planning and reasoning.

Intrigued by the claims of emergent reasoning capabilities in LLMs trained on general web corpora, in this paper, we set out to investigate their planning capabilities. We aim to evaluate (1) the effectiveness of LLMs in generating plans autonomously in commonsense planning tasks and (2) the potential of LLMs as a source of heuristic guidance for other agents (AI planners) in their planning tasks. We conduct a systematic study by generating a suite of instances on domains similar to the ones employed in the International Planning Competition and evaluate LLMs in two distinct modes: autonomous and heuristic. Our findings reveal that LLMs’ ability to generate executable plans autonomously is rather limited, with the best model (GPT-4) having an average success rate of ~12% across the domains. However, the results in the heuristic mode show more promise. In the heuristic mode, we demonstrate that LLM-generated plans can improve the search process for underlying sound planners and additionally show that external verifiers can help provide feedback on the generated plans and back-prompt the LLM for better plan generation.

Despite the surprising power of many modern AI systems that often learn their own representations, there is significant discontent about their inscrutability and the attendant problems in their ability to interact with humans. While alternatives such as neuro-symbolic approaches have been proposed, there is a lack of consensus on what they are about. There are often two independent motivations (i) symbols as a lingua franca for human-AI interaction and (ii) symbols as (system-produced) abstractions use in its internal reasoning. The jury is still out on whether AI systems will need to use symbols in their internal reasoning to achieve general intelligence capabilities. Whatever the answer there is, the need for (human-understandable) symbols in human-AI interaction seems quite compelling. Symbols, like emotions, may well not be sine qua non for intelligence per se, but they will be crucial for AI systems to interact with us humans--as we can neither turn off our emotions not get by without our symbols. In particular, in many human-designed domains, humans would be interested in providing explicit (symbolic) knowledge and advice--and expect machine explanations in kind. This alone requires AI systems to at least do their I/O in symbolic terms. In this blue sky paper, we argue this point of view, and discuss research directions that need to be pursued to allow for this type of human-AI interaction.

In this work, we investigate and model how human trust affects monitoring. We present a web-based human subject study in which the robot is a worker and the human plays the role of a supervisor. First, we evaluate the correlation between the human trust and monitoring by using statistical tests, and then we learn probabilistic models of the behavioral data collected through our user studies. These models can provide us with the likelihood of a human user monitoring a system given their level of trust. Such models can be leveraged in many systems including the ones designed to be resilient to automation bias and complacency.

Creating reinforcement learning (RL) agents that are capable of accepting and leveraging task-specific knowledge from humans has been long identified as a possible strategy for developing scalable approaches for solving long-horizon problems. While previous works have looked at the possibility of using symbolic models along with RL approaches, they tend to assume that the high-level action models are executable at low level and the fluents can exclusively characterize all desirable MDP states. Symbolic models of real world tasks are however often incomplete. To this end, we introduce Approximate Symbolic-Model Guided Reinforcement Learning, wherein we will formalize the relationship between the symbolic model and the underlying MDP that will allow us to characterize the incompleteness of the symbolic model. We will use these models to extract high-level landmarks that will be used to decompose the task. At the low level, we learn a set of diverse policies for each possible task subgoal identified by the landmark, which are then stitched together. We evaluate our system by testing on three different benchmark domains and show how even with incomplete symbolic model information, our approach is able to discover the task structure and efficiently guide the RL agent towards the goal.

Model-reconciliation explanation is a popular framework for generating explanations for planning problems. While the framework has been extended to multiple settings since its introduction for classical planning problems, there is little agreement on the computational complexity of generating minimal model reconciliation explanations in the basic setting. In this paper, we address this lacuna by introducing a decision-version of the model-reconciliation explanation generation problem and we show that it is Sigma-2-P Complete.

Automated Planning techniques can be leveraged to build effective decision support systems that assist the human-in-the-loop. Such systems must provide intuitive explanations when the suggestions made by these systems seem inexplicable to the human. In this regard, we consider scenarios where the user questions the system's suggestion by providing alternatives (referred to as foils). In response, we empower existing decision support technologies to engage in an interactive explanatory dialogue with the user and provide contrastive explanations based on user-specified foils to reach a consensus on proposed decisions. To provide contrastive explanations, we adapt existing techniques in Explainable AI Planning (XAIP). Furthermore, we use this dialog to elicit the user's latent preferences and propose three modes of interaction that use these preferences to provide revised plan suggestions. Finally, we showcase a decision support system that provides all these capabilities.

Existing approaches for generating human-aware agent behaviors have considered different measures of interpretability in isolation. Further, these measures have been studied under differing assumptions, thus precluding the possibility of designing a single framework that captures these measures under the same assumptions. In this paper, we present a unifying Bayesian framework that models a human observer's evolving beliefs about an agent and thereby define the problem of Generalized Human-Aware Planning. We will show that the definitions of interpretability measures like explicability, legibility and predictability from the prior literature fall out as special cases of our general framework. Through this framework, we also bring a previously ignored fact to light that the human-robot interactions are in effect open-world problems, particularly as a result of modeling the human's beliefs over the agent. Since the human may not only hold beliefs unknown to the agent but may also form new hypotheses about the agent when presented with novel or unexpected behaviors.

There is a growing interest in designing robots that can work alongside humans. Such robots will undoubtedly be expected to explain their behavior and decisions. While generating explanations is an actively researched topic, most works tend to focus on methods that generate explanations that are one size fits all. As in the specifics of the user-model are completely ignored. The handful of works that look at tailoring their explanation to the user’s background rely on having specific models of the users (either analytic models or learned labeling models). The goal of this work is thus to propose an end-to-end adaptive explanation generation system that begins by learning the different types of users that the robot could interact with. Then during the interaction with the target user, it is tasked with identifying the type on the fly and adjust its explanations accordingly. The former is achieved by a data-driven clustering approach while for the latter, we compile our explanation generation problem into a POMDP. We demonstrate the usefulness of our system on two domains using state-of-the-art POMDP solvers. We also report the results of a user study that investigates the benefits of providing personalized explanations in a human-robot interaction setting.

Human explanation (e.g., in terms of feature importance) has been recently used to extend the communication channel between human and agent in interactive machine learning. Under this setting, human trainers provide not only the ground truth but also some form of explanation. However, this kind of human guidance was only investigated in supervised learning tasks, and it remains unclear how to best incorporate this type of human knowledge into deep reinforcement learning. In this paper, we present the first study of using human visual explanations in human-in-the-loop reinforcement learning (HIRL). We focus on the task of learning from feedback, in which the human trainer not only gives binary evaluative "good" or "bad" feedback for queried state-action pairs, but also provides a visual explanation by annotating relevant features in images. We propose EXPAND (EXPlanation AugmeNted feeDback) to encourage the model to encode task-relevant features through a context-aware data augmentation that only perturbs irrelevant features in human salient information. We choose five tasks, namely Pixel-Taxi and four Atari games, to evaluate the performance and sample efficiency of this approach. We show that our method significantly outperforms methods leveraging human explanation that are adapted from supervised learning, and Human-in-the-loop RL baselines that only utilize evaluative feedback.

In this work, we present a new planning formalism called Expectation-Aware planning for decision making with humans in the loop where the human's expectations about an agent may differ from the agent's own model. We show how this formulation allows agents to not only leverage existing strategies for handling model differences like explanations (Chakraborti et al. 2017) and explicability (Kulkarni et al. 2019), but can also exhibit novel behaviors that are generated through the combination of these different strategies. Our formulation also reveals a deep connection to existing approaches in epistemic planning. Specifically, we show how we can leverage classical planning compilations for epistemic planning to solve Expectation-Aware planning problems. To the best of our knowledge, the proposed formulation is the first complete solution to planning with diverging user expectations that is amenable to a classical planning compilation while successfully combining previous works on explanation and explicability. We empirically show how our approach provides a computational advantage over our earlier approaches that rely on search in the space of models.

In this paper, we provide a comprehensive outline of the different threads of work in Explainable AI Planning (XAIP) that has emerged as a focus area in the last couple of years and contrast that with earlier efforts in the field in terms of techniques, target users, and delivery mechanisms. We hope that the survey will provide guidance to new researchers in automated planning towards the role of explanations in the effective design of human-in-the-loop systems, as well as provide the established researcher with some perspective on the evolution of the exciting world of explainable planning.

Designing robots capable of generating interpretable behavior is essential for effective human-robot collaboration. This requires robots to be able to generate behavior that aligns with human expectations but exhibiting such behavior in arbitrary environments could be quite expensive for robots, and in some cases, the robot may not even be able to exhibit expected behavior. However, in structured environments (like warehouses, restaurants, etc.), it may be possible to design the environment so as to boost the interpretability of a robot's behavior or to shape the human's expectations of the robot's behavior. In this paper, we investigate the opportunities and limitations of environment design as a tool to promote a particular type of interpretable behavior - known in the literature as explicable behavior. We formulate a novel environment design framework that considers design over multiple tasks and over a time horizon. In addition, we explore the longitudinal effect of explicable behavior and the trade-off that arises between the cost of design and the cost of generating explicable behavior over an extended time horizon.

Proactive Decision Support aims at improving the decision making experience of human decision-makers by enhancing the quality of the decisions and the ease of making them. Given that AI techniques are efficient in searching over a potentially large solution space (of decision) and finding good solutions, it can be used for human-in-the-loop scenarios such as disaster response that demand naturalistic decision making. A human decision-maker, in such scenarios, may experience high-cognitive overload leading to a loss of situational awareness. In this paper, we propose the use of automated task-planning techniques coupled with design principles laid out in the Human-Computer Interaction (HCI) community for developing a proactive decision support system. To this extent, we highlight the capabilities of such a system RADAR and briefly, describe how automated planning techniques help us in providing the varying degrees of assistance. To evaluate the effectiveness of the different capabilities, we conduct ablation studies with human subjects on a synthetic environment for making an interactive plan of study. We found that planning techniques like plan validation and suggestions help to reduce planning time (objective metrics) and improves user satisfaction (subjective metrics) compared to expert human planners without any support.

Plan recognition aims to discover target plans (i.e., sequences of actions) behind observed actions, with history plan libraries or action models in hand. Previous approaches either discover plans by maximally “matching” observed actions to plan libraries, assuming target plans are from plan libraries, or infer plans by executing action models to best explain the observed actions, assuming that complete action models are available. In real-world applications, however, target plans are often not from plan libraries, and complete action models are often not available, since building complete sets of plans and complete action models are often difficult or expensive. In this article, we view plan libraries as corpora and learn vector representations of actions using the corpora; we then discover target plans based on the vector representations. Specifically, we propose two approaches, DUP and RNNPlanner, to discover target plans based on vector representations of actions. DUP explores the EM-style (Expectation Maximization) framework to capture local contexts of actions and discover target plans by optimizing the probability of target plans, while RNNPlanner aims to leverage long-short term contexts of actions based on RNNs (Recurrent Neural Networks) framework to help recognize target plans. In the experiments, we empirically show that our approaches are capable of discovering underlying plans that are not from plan libraries without requiring action models provided. We demonstrate the effectiveness of our approaches by comparing its performance to traditional plan recognition approaches in three planning domains. We also compare DUP and RNNPlanner to see their advantages and disadvantages.

Users of AI systems may rely upon them to produce plans for achieving desired objectives. Such AI systems should be able to compute obfuscated plans whose execution in adversarial situations protects privacy, as well as legible plans which are easy for team members to understand in cooperative situations. We develop a unified framework that addresses these dual problems by computing plans with a desired level of comprehensibility from the point of view of a partially informed observer. For adversarial settings, our approach produces obfuscated plans with observations that are consistent with at least k goals from a set of decoy goals. By slightly varying our framework, we present an approach for producing legible plans in cooperative settings such that the observation sequence projected by the plan is consistent with at most j goals from a set of confounding goals. In addition, we show how the observability of the observer can be controlled to either obfuscate or convey the actions in a plan when the goal is known to the observer. We present theoretical results on the complexity analysis of our approach. We also present an empirical evaluation to show the feasibility and usefulness of our approaches using IPC domains.

The ability of an AI agent to build mental models can open up pathways for manipulating and exploiting the human in the hopes of achieving some greater good. In fact, such behavior does not necessarily require any malicious intent but can rather be borne out of cooperative scenarios. It is also beyond the scope of misinterpretation of intents, as in the case of value alignment problems, and thus can be effectively engineered if desired (i.e. algorithms exist that can optimize such behavior not because models were misspecified but because they were misused). Such techniques pose several unresolved ethical and moral questions with regards to the design of autonomy. In this paper, we illustrate some of these issues in a teaming scenario and investigate how they are perceived by participants in a thought experiment. Finally, we end with a discussion on the moral implications of such behavior from the perspective of the doctor-patient relationship.

Recent work in explanation generation for decision making agents has looked at how unexplained behavior of autonomous systems can be understood in terms of differences in the model of the system and the human's understanding of the same, and how the explanation process as a result of this mismatch can be then seen as a process of reconciliation of these models. Existing algorithms in such settings, while having been built on contrastive, selective and social properties of explanations as studied extensively in the psychology literature, have not, to the best of our knowledge, been evaluated in settings with actual humans in the loop. As such, the applicability of such explanations to human-AI and human-robot interactions remains suspect. In this paper, we set out to evaluate these explanation generation algorithms in a series of studies in a mock search and rescue scenario with an internal semi-autonomous robot and an external human commander. During that process, we hope to demonstrate to what extent the properties of these algorithms hold as they are evaluated by humans.

Recent work has formalized the explanation process in the context of automated planning as one of model reconciliation - i.e. a process by which the planning agent can bring the explainee's (possibly faulty) model of a planning problem closer to its understanding of the ground truth until both agree that its plan is the best possible. The content of explanations can thus range from misunderstandings about the agent's beliefs (state), desires (goals) and capabilities (action model). Though existing literature has considered different kinds of these model differences to be equivalent, literature on the explanations in social sciences has suggested that explanations with similar logical properties may often be perceived differently by humans. In this brief report, we explore to what extent humans attribute importance to different kinds of model differences that have been traditionally considered equivalent in the model reconciliation setting. Our results suggest that people prefer the explanations which are related to the effects of actions.

Human-aware planning involves generating plans that are explicable as well as providing explanations when such plans cannot be found. In this paper, we bring these two concepts together and show how an agent can achieve a trade-off between these two competing characteristics of a plan. In order to achieve this, we conceive a first of its kind planner MEGA that can augment the possibility of explaining a plan in the plan generation process itself. We situate our discussion in the context of recent work on explicable planning and explanation generation and illustrate these concepts in two well-known planning domains, as well as in a demonstration of a robot in a typical search and reconnaissance task. Human factor studies in the latter highlight the usefulness of the proposed approach.

Designing agents capable of explaining complex sequential decisions remains a significant open problem in human-AI interaction. Recently, there has been a lot of interest in developing approaches for generating such explanations for various decision-making paradigms. One such approach has been the idea of explanation as model-reconciliation. The framework hypothesizes that one of the common reasons for a user's confusion could be the mismatch between the user's model of the agent's task model and the model used by the agent to generate the decisions. While this is a general framework, most works that have been explicitly built on this explanatory philosophy have focused on classical planning settings where the model of user's knowledge is available in a declarative form. Our goal in this paper is to adapt the model reconciliation approach to a more general planning paradigm and discuss how such methods could be used when user models are no longer explicitly available. Specifically, we present a simple and easy to learn labeling model that can help an explainer decide what information could help achieve model reconciliation between the user and the agent with in the context of planning with MDPs.

Explainable planning is widely accepted as a prerequisite for autonomous agents to successfully work with humans. While there has been a lot of research on generating explanations of solutions to planning problems, explaining the absence of solutions remains an open and under-studied problem, even though such situations can be the hardest to understand or debug. In this paper, we show that hierarchical abstractions can be used to efficiently generate reasons for unsolvability of planning problems. In contrast to related work on computing certificates of unsolvability, we show that these methods can generate compact, human-understandable reasons for unsolvability. Empirical analysis and user studies show the validity of our methods as well as their computational efficacy on a number of benchmark planning domains.

The recent breakthroughs in Artificial Intelligence (AI) have allowed individuals to rely on automated systems for a variety of reasons. Some of these systems are the currently popular voice-enabled systems like Echo by Amazon and Home by Google that are also called as Intelligent Personal Assistants (IPAs). Though there are rising concerns about privacy and ethical implications, users of these IPAs seem to continue using these systems. We aim to investigate to what extent users are concerned about privacy and how they are handling these concerns while using the IPAs. By utilizing the reviews posted online along with the responses to a survey, this paper provides a set of insights about the detected markers related to user interests and privacy challenges. The insights suggest that users of these systems irrespective of their concerns about privacy, are generally positive in terms of utilizing IPAs in their everyday lives. However, there is a significant percentage of users who are concerned about privacy and take further actions to address related concerns. Some percentage of users expressed that they do not have any privacy concerns but when they learned about the "always listening" feature of these devices, their concern about privacy increased.

Extracting action sequences from texts is challenging, as it requires commonsense inferences based on world knowledge. Although there has been work on extracting action scripts, instructions, navigation actions, etc., they require either the set of candidate actions be provided in advance, or action descriptions are restricted to a specific form, e.g., description templates. In this paper we aim to extract action sequences from texts in \emph{free} natural language, i.e., without any restricted templates, provided the set of actions is unknown. We propose to extract action sequences from texts based on the deep reinforcement learning framework. Specifically, we view ``selecting'' or ``eliminating'' words from texts as ``actions'', and texts associated with actions as ``states''. We build Q-networks to learn policies of extracting actions and extract plans from the labeled texts. We demonstrate the effectiveness of our approach on several datasets with comparison to state-of-the-art approaches.

There is a growing interest within the AI research community in developing autonomous systems capable of explaining their behavior to users. However, the problem of computing explanations for users of different levels of expertise has received little research attention. We propose an approach for addressing this problem by representing the user's understanding of the task as an abstraction of the domain model that the planner uses. We present algorithms for generating minimal explanations in cases where this abstract human model is not known. We reduce the problem of generating an explanation to a search over the space of abstract models and show that while the complete problem is NP-hard, a greedy algorithm can provide good approximations of the optimal solution. We also empirically show that our approach can efficiently compute explanations for a variety of problems.

Recent advances in mixed-reality technologies have renewed interest in alternative modes of communication for human-robot interaction. However, most of the work in this direction has been confined to tasks such as teleoperation, simulation or explication of individual actions of a robot. In this paper, we will discuss how the capability to project intentions affect the task planning capabilities of a robot. Specifically, we will start with a discussion on how projection actions can be used to reveal information regarding the future intentions of the robot at the time of task execution. We will then pose a new planning paradigm - projection-aware planning - whereby a robot can trade off its plan cost with its ability to reveal its intentions using its projection actions. We will demonstrate each of these scenarios with the help of a joint human-robot activity using the HoloLens.

Intelligent robots and machines are becoming pervasive in human populated environments. A desirable capability of these agents is to respond to goal-oriented commands by autonomously constructing task plans. However, such autonomy can add significant cognitive load and potentially introduce safety risks to humans when agents behave in unexpected ways. Hence, for such agents to be helpful, one important requirement is for them to synthesize plans that can be easily understood by humans. While there exists previous work that studied socially acceptable robots that interact with humans in “natural ways”, and work that investigated legible motion planning, there is no general solution for high level task planning. To address this issue, we introduce the notions of plan explicability and predictability. To compute these measures, first, we postulate that humans understand agent plans by associating abstract tasks with agent actions, which can be considered as a labeling process. We learn the labeling scheme of humans for agent plans from training examples using conditional random fields (CRFs). Then, we use the learned model to label a new plan to compute its explicability and predictability. These measures can be used by agents to proactively choose or directly synthesize plans that are more explicable and predictable to humans. We provide evaluations on a synthetic domain and with a physical robot to demonstrate the effectiveness of our approach.

When AI systems interact with humans in the loop, they are often called on to provide explanations for their plans and behavior. Past work on plan explanations primarily involved the AI system explaining the correctness of its plan and the rationale for its decision in terms of its own model. Such soliloquy is wholly inadequate in most realistic scenarios where the humans have domain and task models that differ significantly from that used by the AI system. We posit that the explanations are best studied in light of these differing models. In particular, we show how explanation can be seen as a "model reconciliation problem" (MRP), where the AI system in effect suggests changes to the human's model, so as to make its plan be optimal with respect to that changed human model. We will study the properties of such explanations, present algorithms for automatically computing them, and evaluate the performance of the algorithms.

Finding diverse solutions has become important in many combinatorial search domains, including Automated Planning, Path Planning and Constraint Programming. Much of the work in these directions has however focussed on coming up with appropriate diversity metrics and compiling those metrics in to the solvers/planners. Most approaches use linear-time greedy algorithms for exploring the state space of solution combinations for generating a diverse set of solutions, limiting not only their completeness but also their effectiveness within a time bound. In this paper, we take a combinatorial search perspective on generating diverse solutions. We present a generic bi-level optimization framework for finding cost-sensitive diverse solutions. We propose complete methods under this framework, which guarantee finding a set of cost sensitive diverse solutions satisficing the given criteria whenever there exists such a set. We identify various aspects that affect the performance of these exhaustive algorithms and propose techniques to improve them. Experimental results show the efficacy of the proposed framework compared to an existing greedy approach.

In this brief abstract, we develop a computationally simpler version of the operator count heuristic for a particular class of domains. The contribution of this abstract is thus threefold, we (1) propose an efficient closed form approximation to the operator count heuristic; (2) leverage compressed sensing techniques to obtain an integer approximation in polynomial time; and (3) discuss the relationship of the proposed formulation to existing heuristics and investigate properties of domains where such approaches are useful.

Plan synthesis often requires complete domain models and initial states as input. In many real world applications, it is difficult to build domain models and provide complete initial state beforehand. In this paper we propose to turn to the crowd for help before planning. We assume there are annotators available to provide information needed for building domain models and initial states. However, there might be a substantial amount of discrepancy within the inputs from the crowd. It is thus challenging to address the planning problem with possibly noisy information provided by the crowd. We address the problem by two phases. We first build a set of Human Intelligence Tasks (HITs), and collect values from the crowd. We then estimate the actual values of variables and feed the values to a planner to solve the problem.

Recently there has been a lot of focus on human robot co-habitation issues that are often orthogonal to many aspects of human-robot teaming; e.g. on producing socially acceptable behaviors of robots and de-conflicting plans of robots and humans in shared environments. However, an interesting offshoot of these settings that has largely been overlooked is the problem of planning for serendipity - i.e. planning for stigmergic collaboration without explicit commitments on agents in co-habitation. In this paper we formalize this notion of planning for serendipity for the first time, and provide an Integer Programming based solution for this problem. Further, we illustrate the different modes of this planning technique on a typical Urban Search and Rescue scenario and show a real-life implementation of the ideas on the Nao Robot interacting with a human colleague.

It has long been assumed that for effective human-robot teaming, it is desirable for assistive robots to infer the goals and intents of the humans, and take proactive actions to help them achieve their goals. However, there has not been any systematic evaluation of the accuracy of this claim. On the face of it, there are several ways a proactive robot assistant can in fact reduce the effectiveness of teaming. For example, it can increase the cognitive load of the human teammate by performing actions that are unanticipated by the human. In such cases, even though the teaming performance could be improved, it is unclear whether humans are willing to adapt to robot actions or are able to adapt in a timely manner. Furthermore, misinterpretations and delays in goal and intent recognition due to partial observations and limited communication can also reduce the performance. In this paper, our aim is to perform an analysis of human factors on the effectiveness of such proactive support in human-robot teaming. We perform our evaluation in a simulated Urban Search and Rescue (USAR) task, in which the efficacy of teaming is not only dependent on individual performance but also on teammates' interactions with each other. In this task, the human teammate is remotely controlling a robot while working with an intelligent robot teammate `Mary'. Our main result shows that the subjects generally preferred Mary with the ability to provide proactive support (compared to Mary without this ability). Our results also show that human cognitive load was increased with a proactive assistant (albeit not significantly) even though the subjects appeared to interact with it less.

One subclass of human computation applications are those directed at tasks that involve planning (e.g. tour planning) and scheduling (e.g. conference scheduling). Interestingly, work on these systems shows that even primitive forms of automated oversight on the human contributors helps in significantly improving the effectiveness of the humans/crowd. In this paper, we argue that the automated oversight used in these systems can be viewed as a primitive automated planner, and that there are several opportunities for more sophisticated automated planning in effectively steering the crowd. Straightforward adaptation of current planning technology is however hampered by the mismatch between the capabilities of human workers and automated planners. We identify and partially address two important challenges that need to be overcome before such adaptation of planning technology can occur: (i) interpreting inputs of the human workers (and the requester) and (ii) steering or critiquing plans produced by the human workers, armed only with incomplete domain and preference models. To these ends, we describe the implementation of AI-MIX, a tour plan generation system that uses automated checks and alerts to improve the quality of plans created by human workers; and present a preliminary evaluation of the effectiveness of steering provided by automated planning.

One of the more important aims of graduate artificial intelligence courses is to prepare graduate students to critically evaluate the current literature. The established approaches for this include either asking a student to present a paper in class, or to have the entire class read and discuss a paper. However, neither of these approaches presents incentives for student participation beyond the posting of a single summary or review. In this paper, we describe a class project that uses the popular Easychair conference management system as a pedagogical tool to enable engagement in the peer review process. We report on the deployment of this project in a mediumsized graduate AI class, and present the results of this deployment. We hope that the success of this project in engaging students in the peer review process can be used better train and bolster the future corps of AI reviewers.

Recent efforts in data cleaning of structured data have focused exclusively on problems like data deduplication, record matching, and data standardization; none of these focus on fixing incorrect attribute values in tuples. Correcting values in tuples is typically performed by a minimum cost repair of tuples that violate static constraints like CFDs (which have to be provided by domain experts, or learned from a clean sample of the database). In this paper, we provide a method for correcting individual attribute values in a structured database using a Bayesian generative model and a statistical error model learned from the noisy database directly. We thus avoid the necessity for a domain expert or clean master data. We also show how to efficiently perform consistent query answering using this model over a dirty database, in case write permissions to the database are unavailable. We evaluate our methods over both synthetic and real data.

Human computation applications that involve planning and scheduling are gaining popularity, and the existing literature on such systems shows that any automated oversight on human contributors improves the effectiveness of the crowd. In this paper, we present our ongoing work on the AI-MIX system, which is a first step towards using an automated planning and scheduling system in a crowdsourced planning application. In order to address the mismatch between the capabilities of the crowd and the automated planner, we identify two major challenges -- interpretation, and steering. We also present preliminary empirical results over the tour planning domain, and show how using an automated planner can help improve the quality of plans.

Although many approaches have been developed to form robot coalitions that can achieve a multirobot task, no general methods exist to execute these coalitions, especially when the coordination among the robots is tightly coupled. In this paper, we propose a coordination mechanism as the first step to address coalition execution; it provides a flexible method to reason about synergies with overlapping coalitions (thus enabling multi-tasking robots in multi-robot tasks), which not only improves efficiency, but also reduces resource requirements in task execution. This means that our approach enables tasks that cannot be easily handled before, especially when critical resources are rare but commonly required. Our approach is based on the concept of sensor constraint, which is introduced by the tight coupling (e.g., information sharing) between the robots. We show that our algorithm is sound and complete in finding a coordination solution given a few assumptions, and discuss a distributed implementation. Simulation results are provided to demonstrate the capabilities of this new approach.

Beliefs play an important role in human-robot teaming scenarios, where the robots must reason about other agents' intentions and beliefs in order to inform their own plan generation process, and to successfully coordinate plans with the other agents. In this paper, we cast the evolving and complex structure of beliefs, and inference over them, as a planning and plan recognition problem. We use agent beliefs and intentions modeled in terms of predicates in order to create an automated planning problem instance, which is then used along with a known and complete domain model in order to predict the plan of the agent whose beliefs are being modeled. Information extracted from this predicted plan is used to inform the planning process of the modeling agent, to enable coordination. We also look at an extension of this problem to a plan recognition problem. We conclude by presenting an evaluation of our technique through a case study implemented on a real robot.

We introduce an algorithm to automatically learn probabilistic hierarchical task networks (pHTNs) that capture a user's preferences on plans by observing only the user's behavior. HTNs are a common choice of representation for a variety of purposes in planning, including work on learning in planning. Our contributions are twofold. First, in contrast with prior work, which employs HTNs to represent domain physics or search control knowledge, we use HTNs to model user preferences. Second, while most prior work on HTN learning requires additional information (e.g., annotated traces or tasks) to assist the learning process, our system only takes plan traces as input. Initially, we will assume that users carry out preferred plans more frequently, and thus the observed distribution of plans is an accurate representation of user preference. We then generalize to the situation where feasibility constraints frequently prevent the execution of preferred plans. Taking the prevalent perspective of viewing HTNs as grammars over primitive actions, we adapt an expectation-maximization (EM) technique from the discipline of probabilistic grammar induction to acquire probabilistic context-free grammars (pCFG) that capture the distribution on plans. To account for the difference between the distributions of possible and preferred plans, we subsequently modify this core EM technique by rescaling its input. We empirically demonstrate that the proposed approaches are able to learn HTNs representing user preferences better than the inside-outside algorithm. Furthermore, when feasibility constraints are obfuscated, the algorithm with rescaled input performs better than the algorithm with the original input.

There is increasing awareness in the planning community that depending on complete models impedes the applicability of planning technology in many real world domains where the burden of specifying complete domain models is too high. In this paper, we consider a novel solution for this challenge that combines generative planning on incomplete domain models with a library of plan cases that are known to be correct. While this was arguably the original motivation for case-based planning, most existing case-based planners assume (and depend on) from-scratch planners that work on complete domain models. In contrast, our approach views the plan generated with respect to the incomplete model as a ``skeletal plan'' and augments it with directed mining of plan fragments from library cases. We will present the details of our approach and present an empirical evaluation of our method in comparison to a state-of-the-art case-based planner that depends on complete domain models.

The increasing popularity of Twitter renders improved trust- worthiness and relevance assessment of tweets much more important for search. However, given the limitations on the size of tweets, it is hard to extract measures for ranking from the tweets? content alone. We present a novel ranking method called RAProp, which combines two orthogonal measures of relevance and trustworthiness of a tweet. The first, called Feature Score, measures the trustworthiness of the source of the tweet by extracting features from a 3-layer Twitter ecosystem consisting of users, tweets and webpages. The second measure, called agreement analysis, estimates the trustworthiness of the content of a tweet by analyzing whether the content is independently corroborated by other tweets. We view the candidate result set of tweets as the vertices of a graph, with the edges measuring the estimated agreement between each pair of tweets. The feature score is propagated over this agreement graph to compute the top-k tweets that have both trustworthy sources and independent corroboration. The evaluation of our method on 16 million tweets from the TREC 2011 Microblog Dataset shows that for top-30 precision, we achieve 53% better precision than the current best performing method on the data set, and an improvement of 300% over current Twitter Search.

Most current planners assume complete domain models and focus on generating correct plans. Unfortunately, domain modeling is a laborious and error-prone task, thus real world agents have to plan with incomplete domain models. While domain experts cannot guarantee completeness, often they are able to circumscribe the incompleteness of the model by providing annotations as to which parts of the domain model may be incomplete. In such cases, the goal should be to synthesize plans that are robust with respect to any known incompleteness of the domain. In this paper, we first introduce annotations expressing the knowledge of the domain incompleteness and formalize the notion of plan robustness with respect to an incomplete domain model. We then show an approach to compiling the problem of finding robust plans to the conformant probabilistic planning problem, and present experimental results with Probabilistic-FF planner.

Deep web search engines face the formidable challenge of retrieving high-quality results from the vast collection of searchable databases. Deep web search is a two-step process of selecting the high-quality sources and ranking the results from the selected sources. Though there are existing methods for both the steps, they assess the relevance of the sources and the results using the query-result similarity. When applied to the deep web these methods have two deficiencies. First is that they are agnostic to the correctness (trustworthiness) of the results. Second, the query-based relevance does not consider the importance of the results and sources. These two considerations are essential for the deep web and open collections in general. Since a number of deep web sources provide answers to any query, we conjuncture that the agreements between these answers are helpful in assessing the importance and the trustworthiness of the sources and the results. For assessing source quality, we compute the agreement between the sources as the agreement of the answers returned. While computing the agreement, we also measure and compensate for the possible collusion between the sources. This adjusted agreement is modeled as a graph with sources at the vertices. On this agreement graph, a quality score of a source, that we call SourceRank , is calculated as the stationary visit probability of a random walk. For ranking results, we analyze the second-order agreement between the results. Further extending SourceRank to multidomain search, we propose a source ranking sensitive to the query domains. Multiple domain-specific rankings of a source are computed, and these ranks are combined for the final ranking. We perform extensive evaluations on online and hundreds of Google Base sources spanning across domains. The proposed result and source rankings are implemented in the deep web search engine Factal . We demonstrate that the agreement analysis tracks source corruption. Further, our relevance evaluations show that our methods improve precision significantly over Google Base and the other baseline methods. The result ranking and the domain-specific source ranking are evaluated separately.

During broadcast events such as the Superbowl, the U.S. Presidential and Primary debates, etc., Twitter has become the de facto platform for crowds to share perspectives and commentaries about them. Given an event and an associated large-scale collection of tweets, there are two fundamental research problems that have been receiving increasing attention in recent years. One is to extract the topics covered by the event and the tweets; the other is to segment the event. So far these problems have been viewed separately and studied in isolation. In this work, we argue that these problems are in fact inter-dependent and should be addressed together. We develop a joint Bayesian model that performs topic modeling and event segmentation in one unified framework. We evaluate the proposed model both quantitatively and qualitatively on two large-scale tweet datasets associated with two events from different domains to show that it improves significantly over baseline models.

Robots are currently being used in and developed for critical HRI applications such as search and rescue. In these scenarios, humans operating under changeable and high-stress conditions must communicate effectively with autonomous agents, necessitating that such agents be able to respond quickly and effectively to rapidly-changing conditions and expectations. We demonstrate a robot planner that is able to utilize new information, specifically information originating in spoken input produced by human operators.

Multi-Agent Plan Recognition (MAPR) aims to recognize dynamic team structures and team behaviors from the observed team traces (activity sequences) of a set of intelligent agents. Previous MAPR approaches required a library of team activity sequences (team plans) be given as input. However, collecting a library of team plans to ensure adequate coverage is often difficult and costly. In this paper, we relax this constraint, so that team plans are not required to be provided beforehand. We assume instead that a set of action models are available. Such models are often already created to describe domain physics; i.e., the preconditions and effects of effects actions. We propose a novel approach for recognizing multi-agent team plans based on such action models rather than libraries of team plans. We encode the resulting MAPR problem as a \emph{satisfiability problem} and solve the problem using a state-of-the-art weighted MAX-SAT solver. Our approach also allows for incompleteness in the observed plan traces. Our empirical studies demonstrate that our algorithm is both effective and efficient in comparison to state-of-the-art MAPR methods based on plan libraries.

We present a novel ensemble architecture for learning problem-solving techniques from a very small number of expert solutions and demonstrate its effectiveness in a complex real-world domain. The key feature of our “Generalized Integrated Learning Architecture” (GILA) is a set of heterogeneous independent learning and reasoning (ILR) components, coordinated by a central meta-reasoning executive (MRE). The ILRs are weakly coupled in the sense that all coordination during learning and performance happens through the MRE. Each ILR learns independently from a small number of expert demonstrations of a complex task. During performance, each ILR proposes partial solutions to subproblems posed by the MRE, which are then selected from and pieced together by the MRE to produce a complete solution. The heterogeneity of the learner-reasoners allows both learning and problem solving to be more effective because their abilities and biases are complementary and synergistic. We describe the application of this novel learning and problem solving architecture to the domain of airspace management, where multiple requests for the use of airspaces need to be deconflicted, reconciled, and managed automatically. Formal evaluations show that our system performs as well as or better than humans after learning from the same training data. Furthermore, GILA outperforms any individual ILR run in isolation, thus demonstrating the power of the ensemble architecture for learning and problem solving.

One immediate challenge in searching the deep web databases is source selection - i.e. selecting the most relevant web databases for answering a given query. The existing database selection methods (both text and relational) assess the source quality based on the query-similarity-based relevance assessment. When applied to the deep web these methods have two deficiencies. First is that the methods are agnostic to the correctness (trustworthiness) of the sources. Secondly, the query based relevance does not consider the importance of the results. These two considerations are essential for the open collections like the deep web. Since a number of sources provide answers to any query, we conjuncture that the agreements between these answers are likely to be helpful in assessing the importance and the trustworthiness of the sources. We compute the agreement between the sources as the agreement of the answers returned. While computing the agreement, we also measure and compensate for possible collusion between the sources. This adjusted agreement is modeled as a graph with sources at the vertices. On this agreement graph, a quality score of a source that we call SourceRank, is calculated as the stationary visit probability of a random walk. We evaluate SourceRank in multiple domains, including sources in Google Base, with sizes up to 675 sources. We demonstrate that the SourceRank tracks source corruption. Further, our relevance evaluations show that SourceRank improves precision by 22-60% over the Google Base and the other baseline methods. SourceRank has been implemented in a system called Factal.

In this paper, we present an integrated planning and robotic architecture that actively directs an agent engaged in an urban search and rescue (USAR) scenario. We describe three salient features that comprise the planning component of this system, namely (1) the ability to plan in a world open with respect to objects, (2) execution monitoring and replanning abilities, and (3) handling soft goals, and detail the interaction of these parts in representing and solving the USAR scenario at hand. We show that though insufficient in an individual capacity, the integration of this trio of features is sufficient to solve the scenario that we present. We test our system with an example problem that involves soft and hard goals, as well as goal deadlines and action costs, and show that the planner is capable of incorporating sensing actions and execution monitoring in order to produce goal-fulfilling plans that maximize the net benefit accrued.

Many web databases can be seen as providing partial and overlapping information about entities in the world. To answer queries effectively, we need to integrate the information about the individual entities that are fragmented over multiple sources. At first blush this is just the inverse of traditional database normalization problem - rather than go from a universal relation to normalized tables, we want to reconstruct the universal relation given the tables (sources). The standard way of reconstructing the entities will involve joining the tables. Unfortunately, because of the autonomous and decentralized way in which the sources are populated, they often do not have Primary Key - Foreign Key relations. While tables do share attributes, naive joins over these shared attributes can result in reconstruction of many spurious entities thus seriously compromising precision. Our system, SMARTINT is aimed at addressing the problem of data integration in such scenarios. Given a query, our system uses the Approximate Functional Dependencies(AFDs) to piece together a tree of relevant tables and schemas for joining them. The result tuples produced by our system are able to strike a favorable balance between precision and recall.

Planning research has returned to the issue of optimizing costs (rather than sizes) of plans. A prevalent perception, at least among non-experts in search, is that graph search for optimizing the size of paths generalizes more or less trivially to optimizing the cost of paths. While this kind of generalization is usually straightforward for graph theorems, graph algorithms are a different story. In particular, implementing a search evaluation function by substituting cost for size is a Bad Idea. Though experts have stated as much, cutting-edge practitioners are still learning of the consequences the hard way; here we mount a forceful indictment on the inherent dangers of cost-based search.

One immediate challenge in searching the deep web databases is source selection - i.e. selecting the most relevant web databases for answering a given query. The existing database selection methods (both text and relational) assess the source quality based on the query-similarity-based relevance assessment. When applied to the deep web these methods have two deficiencies. First is that the methods are agnostic to the correctness (trustworthiness) of the sources. Secondly, the query based relevance does not consider the importance of the results. These two considerations are essential for the open collections like the deep web. Since a number of sources provide answers to any query, we conjuncture that the agreements between these answers are likely to be helpful in assessing the importance and the trustworthiness of the sources. We compute the agreement between the sources as the agreement of the answers returned. While computing the agreement, we also measure and compensate for possible collusion between the sources. This adjusted agreement is modeled as a graph with sources at the vertices. On this agreement graph, a quality score of a source that we call SourceRank, is calculated as the stationary visit probability of a random walk. We evaluate SourceRank in multiple domains, including sources in Google Base, with sizes up to 675 sources. We demonstrate that the SourceRank tracks source corruption. Further, our relevance evaluations show that SourceRank improves precision by 22-60% over the Google Base and the other baseline methods. SourceRank has been implemented in a system called Factal.

As the number of applications for human-robot teaming continue to rise, there is an increasing need for planning technologies that can guide robots in such teaming scenarios. In this article, we focus on adapting planning technology to Urban Search And Rescue (USAR) with a human-robot team. We start by showing that several aspects of state-of-the-art planning technology, including temporal planning, partial satisfaction planning, and replanning, can be gainfully adapted to this scenario. We then note that human-robot teaming also throws up an additional critical challenge, namely, enabling existing planners, which work under closed-world assumptions, to cope with the open worlds that are characteristic of teaming problems such as USAR. In response, we discuss the notion of conditional goals, and describe how we represent and handle a specific class of them called open world quantified goals. Finally, we describe how the planner, and its open world extensions, are integrated into a robot control architecture, and provide an empirical evaluation over USAR experimental runs to establish the effectiveness of the planning components.

Autonomous robots that operate in real-world domains face multiple challenges that make planning and goal selection difficult. Not only must planning and execution occur in real time, newly acquired knowledge can invalidate previous plans, and goals and their utilities can change during plan execution. However, these events can also provide opportunities, if the architecture is designed to react appropriately. We present here an architecture that integrates the SapaReplan planner with the DIARC robot architecture, allowing the architecture to react dynamically to changes in the robot's goal structures.

We represent planning as a set of loosely coupled network flow problems, where each network corresponds to one of the state variables in the planning domain. The network nodes correspond to the state variable values and the network arcs correspond to the value transitions. The planning problem is to find a path (a sequence of actions) in each network such that, when merged, they constitute a feasible plan. In this paper we present a number of integer programming formulations that model these loosely coupled networks with varying degrees of flexibility. Since merging may introduce exponentially many ordering constraints we implement a so-called branch-and-cut algorithm, in which these constraints are dynamically generated and added to the formulation when needed. Our results are very promising, they improve upon previous planning as integer programming approaches and lay the foundation for integer programming approaches for cost optimal planning.

Incompleteness due to missing attribute values (aka "null values") is very common in autonomous Web databases, on which user accesses are usually supported through mediators. Traditional query processing techniques that focus on the strict soundness of answer tuples often ignore tuples with critical missing attributes, even if they wind up being relevant to a user query. Ideally we would like the mediator to retrieve such relevant uncertain answers and gauge their relevance by accessing their likelihood of being relevant answers to the query. However, the autonomous nature of the databases poses several challenges, such as the restricted access privileges, limited query patterns, and sensitivity of database and network resource consumption in the Web environment. We introduce a novel query rewriting and optimization framework QPIAD that tackles these challenges to retrieve relevant uncertain answers. Our technique involves reformulating the user query based on approximate functional dependencies (AFDs) among the database attributes and ranking these queries using value distributions learned from naive Bayes classifiers. Empirical studies demonstrate the effectiveness of our approach in retrieving relevant uncertain answers with high precision, high recall and manageable cost.

Current approaches for answering queries with imprecise constraints require user-specific distance metrics and importance measures for attributes of interest - metrics that are hard to elicit from lay users. We present AIMQ, a domain and user independent approach for answering imprecise queries over autonomous Web databases. We developed methods for query relaxation that use approximate functional dependencies. We also present an approach to automatically estimate the similarity between values of categorical attributes. Experimental results demonstrating the robustness, efficiency and effectiveness of AIMQ are presented. Results of a preliminary user study demonstrating the high precision of the AIMQ system is also provided.

Some recent works in conditional planning have proposed reachability heuristics to improve planner scalability, but many lack a formal description of the properties of their distance estimates. To place previous work in context and extend work on heuristics for conditional planning, we provide a formal basis for distance estimates between belief states. We give a definition for the distance between belief states that relies on aggregating underlying state distance measures. We give several techniques to aggregate state distances and their associated properties. Many existing heuristics exhibit a subset of the properties, but in order to provide a standardized comparison we present several generalizations of planning graph heuristics that are used in a single planner. We compliment our belief state distance estimate framework by also investigating efficient planning graph data structures that incorporate BDDs to compute the most effective heuristics. We developed two planners to serve as test-beds for our investigation. The first, CAltAlt, is a conformant regression planner that uses A* search. The second, POND, is a conditional progression planner that uses AO* search. We show the relative effectiveness of our heuristic techniques within these planners. We also compare the performance of these planners with several state of the art approaches in conditional planning.

The Optiplan planning system is the first integer programming-based planner that successfully participated in the international planning competition. This engineering note describes the architecture of Optiplan and provides the integer programming formulation that enabled it to perform reasonably well in the competition. We also touch upon some recent developments that make integer programming encodings significantly more competitive.

Recent work in data integration has shown the importance of statistical information about the coverage and overlap of sources for efficient query processing. Despite this recognition, there are no effective approaches for learning the needed statistics. The key challenge in learning such statistics is keeping the number of needed statistics low enough to have the storage and learning costs manageable. In this paper, we present a set of connected techniques that estimate the coverage and overlap statistics, while keeping the needed statistics tightly under control. Our approach uses a hierarchical classification of the queries and threshold-based variants of familiar data mining techniques to dynamically decide the level of resolution at which to learn the statistics. We describe the details of our method, and, present experimental results demonstrating the efficiency of the learning algorithms and the effectiveness of the learned statistics over both controlled data sources and in the context of BibFinder with autonomous online sources.

Query optimization in data integration requires source coverage and overlap statistics. Gathering and storing the required statistics presents many challenges, not the least of which is controlling the amount of statistics learned. We introduce StatMiner, a novel statistics mining approach which automatically generates attribute value hierarchies, efficiently discovers frequently accessed query classes based on the learned attribute value hierarchies, and learns statistics only with respect to these classes. We describe the details of our method, and present experimental results demonstrating the efficiency and effectiveness of our approach. Our experiments are done in the context of BibFinder, a publicly fielded bibliography mediator.

Current approaches for answering queries with imprecise constraints require users to provide distance metrics and importance measures for attributes of interest. In this paper we focus on providing a domain and end-user independent solution for supporting imprecise queries over Web databases without affecting the underlying database. We propose a query processing framework that integrates techniques from IR and database research to efficiently determine answers for imprecise queries. We mine and use approximate functional dependencies between attributes to create precise queries having tuples relevant to the given imprecise query. An approach to automatically estimate the semantic distances between values of categorical attributes is also proposed. We provide preliminary results showing the utility of our approach.

SAPA is a domain-independent heuristic forward chaining planner that can handle durative actions, metric resource constraints, and deadline goals. It is designed to be capable of handling the multi-objective nature of metric temporal planning. Our technical contributions include (i) planning-graph based methods for deriving heuristics that are sensitive to both cost and makespan (ii) techniques for adjusting the heuristic estimates to take action interactions and metric resource limitations into account and (iii) a linear time greedy post-processing technique to improve execution flexibility of the solution plans. An implementation of SAPA using many of the techniques presented in this paper was one of the best domain independent planners for domains with metric and temporal constraints in the third International Planning Competition, held at AIPS-02. We describe the technical details of extracting the heuristics and present an empirical evaluation of the current implementation of SAPA.

Despite their near dominance, heuristic state search planners still lag behind disjunctive planners in the generation of parallel plans in classical planning. The reason is that directly searching for parallel solutions in state space planners would require the planners to branch on all possible subsets of parallel actions, thus increasing the branching factor exponentially. We present a variant of our heuristic state search planner AltAlt, called AltAltp which generates parallel plans by using greedy online parallelization of partial plans. The greedy approach is significantly informed by the use of novel distance heuristics that AltAltp derives from a graphplan-style planning graph for the problem. While this approach is not guaranteed to provide optimal parallel plans, empirical results show that AltAltp is capable of generating good quality parallel plans at a fraction of the cost incurred by the disjunctive planners.

Recent work in data integration has shown the importance of statistical information about the coverage and overlap of sources for efficient query processing. Despite this recognition there are no effective approaches for learning the needed statistics. The key challenge in learning such statistics is keeping the number of needed statistics low enough to have the storage and learning costs manageable. Naive approaches can become infeasible very quickly. In this paper we present a set of connected techniques that estimate the coverage and overlap statistics while keeping the needed statistics tightly under control. Our approach uses a hierarchical classification of the queries, and threshold based variants of familiar data mining techniques to dynamically decide the level of resolution at which to learn the statistics. We describe the details of our method, and present experimental results demonstrating the efficiency of the learning algorithms and the effectiveness of the learned statistics.

Existing approaches for optimizing queries in data integration use decoupled strategies--attempting to optimize coverage and cost in two separate phases. Since sources tend to have a variety of access limitations, such phased optimization of cost and coverage can unfortunately lead to expensive planning as well as highly inefficient plans. In this paper we present techniques for joint optimization of cost and coverage of the query plans. Our algorithms search in the space of parallel query plans that support multiple sources for each subgoal conjunct. The refinement of the partial plans takes into account the potential parallelism between source calls, and the binding compatibilities between the sources included in the plan. We start by introducing and motivating our query plan representation. We then briefly review how to compute the cost and coverage of a parallel plan. Next, we provide both a System-R style query optimization algorithm as well as a greedy local search algorithm for searching in the space of such query plans. Finally we present a simulation study that demonstrates that the plans generated by our approach will be significantly better, both in terms of planning cost, and in terms of plan execution cost, compared to the existing approaches.

This paper reviews the connections between Graphplan's planning-graph and the dynamic constraint satisfaction problem and motivates the need for adapting CSP search techniques to the Graphplan algorithm. It then describes how explanation based learning, dependency directed backtracking, dynamic variable ordering, forward checking, sticky values and random-restart search strategies can be adapted to Graphplan. Empirical results are provided to demonstrate that these augmentations improve Graphplan's performance significantly (up to 1000x speedups) on several benchmark problems. Special attention is paid to the explanation-based learning and dependency directed backtracking techniques as they are empirically found to be most useful in improving the performance of Graphplan.

Existing plan synthesis approaches in artificial intelligence fall into two categories -- domain independent and domain dependent. The domain independent approaches are applicable across a variety of domains, but may not be very efficient in any one given domain. The domain dependent approaches need to be (re)designed for each domain separately, but can be very efficient in the domain for which they are designed. One enticing alternative to these approaches is to automatically synthesize domain independent planners given the knowledge about the domain and the theory of planning. In this paper, we investigate the feasibility of using existing automated software synthesis tools to support such synthesis. Specifically, we describe an architecture called CLAY in which the Kestrel Interactive Development System (KIDS) is used to derive a domain-customized planner through a semi-automatic combination of a declarative theory of planning, and the declarative control knowledge specific to a given domain, to semi-automatically combine them to derive domain-customized planners. We discuss what it means to write a declarative theory of planning and control knowledge for KIDS, and illustrate our approach by generating a class of domain-specific planners using state space refinements. Our experiments show that the synthesized planners can outperform classical refinement planners (implemented as instantiations of UCP, Kambhampati & Srivastava, 1995), using the same control knowledge. We will contrast the costs and benefits of the synthesis approach with conventional methods for customizing domain independent planners.

Plan synthesis approaches in AI fall into two categories: domain-independent and domain-dependent. The domain-independent approaches are applicable across a variety of domains, but may not be very efficient in any one given domain. The domain-dependent approaches can be very efficient for the domain for which they are designed, but would need to be written separately for each domain of interest. The tediousness and the error-proneness of manual coding have hither-to inhibited work on domain-dependent planners. In this paper we describe a novel way of automating the development of domain dependent planners using knowledge-based software synthesis tools. Specifically, we describe an architecture called CLAY in which the Kestrel Interactive Development System (KIDS) is used in conjunction with a declarative theory of domain independent planning, and the declarative control knowledge specific to a given domain, to semi-automatically derive customized planning code. We discuss what it means to write declarative theory of planning and control knowledge for KIDS, and illustrate it by generating a range of domain-specific planners using state space and plan space refinements. We demonstrate that the synthesized planners can have superior performance compared to classical refinement planners using the same control knowledge.

Case-Based Planning (CBP) provides a way of scaling up domain-independent planning to solve large problems in complex domains. It replaces the detailed and lengthy search for a solution with the retrieval and adaptation of previous planning experiences. In general, CBP has been demonstrated to improve performance over generative (from-scratch) planning. However, the performance improvements it provides are dependent on adequate judgements as to problem similarity. In particular, although CBP may substantially reduce planning effort overall, it is subject to a mis-retrieval problem. The success of CBP depends on these retrieval errors being relatively rare. This paper describes the design and implementation of a replay framework for the case-based planner DERSNLP+EBL. DERSNLP+EBL extends current CBP methodology by incorporating explanation-based learning techniques that allow it to explain and learn from the retrieval failures it encounters. These techniques are used to refine judgements about case similarity in response to feedback when a wrong decision has been made. The same failure analysis is used in building the case library, through the addition of repairing cases. Large problems are split and stored as single goal subproblems. Multi-goal problems are stored only when these smaller cases fail to be merged into a full solution. An empirical evaluation of this approach demonstrates the advantage of learning from experienced retrieval failure.

The ability to reuse existing plans to solve new planning problems can enable a domain‐independent planner to improve its average case efficiency by exploiting the problem distribution and avoiding repetition of planning effort. The pay‐off from plan reuse, however, crucially depends on finding effective solutions to two important underlying control problems: (i) controlling the retrieval of an appropriate plan and mapping to be reused in a new situation, and (ii) controlling the modification (refitting) of the retrieved plan so as to minimize perturbation to the applicable parts of the plan. This paper is concerned with the development of efficient domain‐independent solutions to these two problems. For the retrieval, it provides a domain independent similarity metric that utilizes the plan causal dependency structure to estimate the utility of reusing a given plan in a new problem situation. For the refitting, it presents a minimum‐conflict heuristic, again based on the causal dependency structure of the plan, to conservatively control the modification. The paper also discusses the implementation and evaluation of these strategies within the PRIAR plan modification framework.

Many real-world planning problems involve substantial amounts of domain-specific reasoning that is either awkward or inefficient to encode in a general purpose planner. A hybrid planning architecture for such domains is proposed. It utilizes a set of specialists to complement both the overall expressiveness and the efficiency of a traditional hierarchical planner. Such an architecture promises to retain the flexibility and generality of a classical planning framework while allowing deeper and more efficient domain-specific reasoning through specialists. The architecture has several ramifications on the internal operations of the planner as well as its interactions with the specialists. Continual interactions between the planner and the specialists necessitate an incremental, interactive, and least-commitment oriented approach to planning. As the planner and the specialists in such a model may use heterogeneous reasoning mechanisms and representations, a complete understanding of the operations of one by the other is not possible.>

The problem of automatic collision-free path planning is central to mobile robot applications. An approach to automatic path planning based on a quadtree representation is presented. Hierarchical path-searching methods are introduced, which make use of this multiresolution representation, to speed up the path planning process considerably. The applicability of this approach to mobile robot path planning is discussed.

The Computer Vision Laboratory at the University of Maryland is designing and developing a vision system for autonomous ground navigation. Our approach to visual navigation segments the task into three levels called long range, intermediate range and short range navigation. At the long range, one would first generate a plan for the day's outing, identifying the starting location, the goal, and a low resolution path for moving from the start to the goal. From time to time, during the course of the outing, one may want to establish his position with respect to the long range plan. This could be accomplished by visually identifying landmarks of known location, and then triangulating to determine current position. We describe a vision system for position determination that we have developed as part of this project. At the intermediate range, one would look ahead to determine generally safe directions of travel called corridors of free space. Short range navigation is the process that, based on a detailed topographic analysis of one's immediate environment, enables us to safely navigate around obstacles in the current corridor of free space along a track of safe passage. We describe a quadtree based path planning algorithm which could serve as the basis for identifying such tracks of safe passage.