The scalability of instructable agents in robotics or gaming is often hindered by limited data that pairs instructions with agent trajectories. However, large datasets of unannotated trajectories containing sequences of various agent behaviour (play trajectories) are often available. In a semi-supervised setup, we explore methods to extract labelled segments from play trajectories. The goal is to augment a small annotated dataset of instruction-trajectory pairs to improve the performance of an instruction-following policy trained downstream via imitation learning. Assuming little variation in segment length, recent video segmentation methods can effectively extract labelled segments. To address the constraint of segment length, we propose Play Segmentation (PS), a probabilistic model that finds maximum likely segmentations of extended subsegments, while only being trained on individual instruction segments. Our results in a game environment and a simulated robotic gripper setting underscore the importance of segmentation; randomly sampled segments diminish performance, while incorporating labelled segments from PS improves policy performance to the level of a policy trained on twice the amount of labelled data.

Exposure bias is a well-known issue in recommender systems where items and suppliers are not equally represented in the recommendation results. This bias becomes particularly problematic over time as a few items are repeatedly over-represented in recommendation lists, leading to a feedback loop that further amplifies this bias. Although extensive research has addressed this issue in model-based or neighborhood-based recommendation algorithms, less attention has been paid to online recommendation models, such as those based on top-K contextual bandits, where recommendation models are dynamically updated with ongoing user feedback. In this paper, we study exposure bias in a class of well-known contextual bandit algorithms known as Linear Cascading Bandits,. We analyze these algorithms in their ability to handle exposure bias and provide a fair representation of items in the recommendation results. Our analysis reveals that these algorithms fail to mitigate exposure bias in the long run during the course of ongoing user interactions. We propose an Exposure-Aware reward model that updates the model parameters based on two factors: 1) implicit user feedback and 2) the position of the item in the recommendation list. The proposed model mitigates exposure bias by controlling the utility assigned to the items based on their exposure in the recommendation list. Our experiments with two real-world datasets show that our proposed reward model improves the exposure fairness of the linear cascading bandits over time while maintaining the recommendation accuracy. It also outperforms the current baselines. Finally, we prove a high probability upper regret bound for our proposed model, providing theoretical guarantees for its performance.

Conventional reinforcement learning (RL) methods can successfully solve a wide range of sequential decision problems. However, learning policies that can generalize predictably across multiple tasks in a setting with non-Markovian reward specifications is a challenging problem. We propose to use successor features to learn a set of local policies that each solves a well-defined subproblem. In a task described by a finite state automaton (FSA) that involves the same set of subproblems, the combination of these local policies can then be used to generate an optimal solution without additional learning. In contrast to other methods that combine local policies via planning, our method asymptotically attains global optimality, even in stochastic environments.

Two typical forms of bias in user interaction data with recommender systems (RSs) are popularity bias and positivity bias, which manifest themselves as the over-representation of interactions with popular items or items that users prefer, respectively. Debiasing methods aim to mitigate the effect of selection bias on the evaluation and optimization of RSs. However, existing debiasing methods only consider single-factor forms of bias, e.g., only the item (popularity) or only the rating value (positivity). This is in stark contrast with the real world where user selections are generally affected by multiple factors at once. In this work, we consider multifactorial selection bias in RSs. Our focus is on selection bias affected by both item and rating value factors, which is a generalization and combination of popularity and positivity bias. While the concept of multifactorial bias is intuitive, it brings a severe practical challenge as it requires substantially more data for accurate bias estimation. As a solution, we propose smoothing and alternating gradient descent techniques to reduce variance and improve the robustness of its optimization. Our experimental results reveal that, with our proposed techniques, multifactorial bias corrections are more effective and robust than single-factor counterparts on real-world and synthetic datasets.

We present Nonparametric Approximation of Inter-Trace returns (NAIT), a Reinforcement Learning algorithm for discrete action, pixel-based environments that is both highly sample and computation efficient. NAIT is a lazy-learning approach with an update that is equivalent to episodic Monte-Carlo on episode completion, but that allows the stable incorporation of rewards while an episode is ongoing. We make use of a fixed domain-agnostic representation, simple distance based exploration and a proximity graph-based lookup to facilitate extremely fast execution. We empirically evaluate NAIT on both the 26 and 57 game variants of ATARI100k where, despite its simplicity, it achieves competitive performance in the online setting with greater than 100x speedup in wall-time.

Reinforcement learning is a promising paradigm for solving sequential decision-making problems, but low data efficiency and weak generalization across tasks are bottlenecks in real-world applications. Model-based meta reinforcement learning addresses these issues by learning dynamics and leveraging knowledge from prior experience. In this paper, we take a closer look at this framework and propose a new posterior sampling based approach that consists of a new model to identify task dynamics together with an amortized policy optimization step. We show that our model, called a graph structured surrogate model (GSSM), achieves competitive dynamics prediction performance with lower model complexity. Moreover, our approach in policy search is able to obtain high returns and allows fast execution by avoiding test-time policy gradient updates.

Enabling reinforcement learning (RL) agents to leverage a knowledge base while learning from experience promises to advance RL in knowledge intensive domains. However, it has proven difficult to leverage knowledge that is not manually tailored to the environment. We propose to use the subclass relationships present in open-source knowledge graphs to abstract away from specific objects. We develop a residual policy gradient method that is able to integrate knowledge across different abstraction levels in the class hierarchy. Our method results in improved sample efficiency and generalisation to unseen objects in commonsense games, but we also investigate failure modes, such as excessive noise in the extracted class knowledge or environments with little class structure.

In robotic tasks, we encounter the unique strengths of (1) reinforcement learning (RL) that can handle high-dimensional observations as well as unknown, complex dynamics and (2) planning that can handle sparse and delayed rewards given a dynamics model. Combining these strengths of RL and planning, we propose the Value Refinement Network (VRN), in this work. Our VRN is an RL-trained neural network architecture that learns to locally refine an initial (value-based) plan in a simplified (2D) problem abstraction based on detailed local sensory observations. We evaluate the VRN on simulated robotic (navigation) tasks and demonstrate that it can successfully refine sub-optimal plans to match the performance of more costly planning in the non-simplified problem. Furthermore, in a dynamic environment, the VRN still enables high task completion without global re-planning.

Hex is a turn-based two-player connection game with a high branching factor, making the game arbitrarily complex with increasing board sizes. As such, top-performing algorithms for playing Hex rely on accurate evaluation of board positions using neural networks. However, the limited interpretability of neural networks is problematic when the user wants to understand the reasoning behind the predictions made. In this paper, we propose to use propositional logic expressions to describe winning and losing board game positions, facilitating precise visual interpretation. We employ a Tsetlin Machine (TM) to learn these expressions from previously played games, describing where pieces must be located or not located for a board position to be strong. Extensive experiments on$6\times 6$boards compare our TM-based solution with popular machine learning algorithms like XGBoost, InterpretML, decision trees, and neural networks, considering various board configurations with 2 to 22 moves played. On average, the TM testing accuracy is 92.1%, outperforming all the other evaluated algorithms. We further demonstrate the global interpretation of the logical expressions, and map them down to particular board game configurations to investigate local interpretability. We believe the resulting interpretability establishes building blocks for accurate assistive AI and human-AI collaboration, also for more complex prediction tasks.

Recent works on machine learning for combinatorial optimization have shown that learning based approaches can outperform heuristic methods in terms of speed and performance. In this paper, we consider the problem of finding an optimal topological order on a directed acyclic graph (DAG) with focus on the memory minimization problem which arises in compilers. We propose an end-to-end machine learning based approach for topological ordering using an encoder-decoder framework. Our encoder is a novel attention based graph neural network architecture called \emph{Topoformer} which uses different topological transforms of a DAG for message passing. The node embeddings produced by the encoder are converted into node priorities which are used by the decoder to generate a probability distribution over topological orders. We train our model on a dataset of synthetically generated graphs called layered graphs. We show that our model outperforms, or is on-par, with several topological ordering baselines while being significantly faster on synthetic graphs with up to 2k nodes. We also train and test our model on a set of real-world computation graphs, showing performance improvements.

Neural processes (NPs) formulate exchangeable stochastic processes and are promising models for meta learning that do not require gradient updates during the testing phase. However, most NP variants place a strong emphasis on a global latent variable. This weakens the approximation power and restricts the scope of applications using NP variants, especially when data generative processes are complicated.To resolve these issues, we propose to combine the Mixture of Expert models with Neural Processes to develop more expressive exchangeable stochastic processes, referred to as Mixture of Expert Neural Processes (MoE-NPs). Then we apply MoE-NPs to both few-shot supervised learning and meta reinforcement learning tasks. Empirical results demonstrate MoE-NPs' strong generalization capability to unseen tasks in these benchmarks.

In policy search methods for reinforcement learning (RL), exploration is often performed by injecting noise either in action space at each step independently or in parameter space over each full trajectory. In prior work, it has been shown that with linear policies, a more balanced trade-off between these two exploration strategies is beneficial. However, that method did not scale to policies using deep neural networks. In this paper, we introduce deep coherent exploration, a general and scalable exploration framework for deep RL algorithms for continuous control, that generalizes step-based and trajectory-based exploration. This framework models the last layer parameters of the policy network as latent variables and uses a recursive inference step within the policy update to handle these latent variables in a scalable manner. We find that deep coherent exploration improves the speed and stability of learning of A2C, PPO, and SAC on several continuous control tasks.

Solving robotic navigation tasks via reinforcement learning (RL) is challenging due to their sparse reward and long decision horizon nature. However, in many navigation tasks, high-level (HL) task representations, like a rough floor plan, are available. Previous work has demonstrated efficient learning by hierarchal approaches consisting of path planning in the HL representation and using sub-goals derived from the plan to guide the RL policy in the source task. However, these approaches usually neglect the complex dynamics and sub-optimal sub-goal-reaching capabilities of the robot during planning. This work overcomes these limitations by proposing a novel hierarchical framework that utilizes a trainable planning policy for the HL representation. Thereby robot capabilities and environment conditions can be learned utilizing collected rollout data. We specifically introduce a planning policy based on value iteration with a learned transition model (VI-RL). In simulated robotic navigation tasks, VI-RL results in consistent strong improvement over vanilla RL, is on par with vanilla hierarchal RL on single layouts but more broadly applicable to multiple layouts, and is on par with trainable HL path planning baselines except for a parking task with difficult non-holonomic dynamics where it shows marked improvements.

Neural processes (NPs) constitute a family of variational approximate models for stochastic processes with promising properties in computational efficiency and uncertainty quantification. These processes use neural networks with latent variable inputs to induce a predictive distribution. However, the expressiveness of vanilla NPs is limited as they only use a global latent variable, while target-specific local variation may be crucial sometimes. To address this challenge, we investigate NPs systematically and present a new variant of NP model that we call Doubly Stochastic Variational Neural Process (DSVNP). This model combines the global latent variable and local latent variables for prediction. We evaluate this model in several experiments, and our results demonstrate competitive prediction performance in multi-output regression and uncertainty estimation in classification.

In today’s clinical practice, magnetic resonance imaging (MRI) is routinely accelerated through subsampling of the associated Fourier domain. Currently, the construction of these subsampling strategies - known as experimental design - relies primarily on heuristics. We propose to learn experimental design strategies for accelerated MRI with policy gradient methods. Unexpectedly, our experiments show that a simple greedy approximation of the objective leads to solutions nearly on-par with the more general non-greedy approach. We offer a partial explanation for this phenomenon rooted in greater variance in the non-greedy objective's gradient estimates, and experimentally verify that this variance hampers non-greedy models in adapting their policies to individual MR images. We empirically show that this adaptivity is key to improving subsampling designs.

This paper introduces MDP homomorphic networks for deep reinforcement learning. MDP homomorphic networks are neural networks that are equivariant under symmetries in the joint state-action space of an MDP. Current approaches to deep reinforcement learning do not usually exploit knowledge about such structure. By building this prior knowledge into policy and value networks using an equivariance constraint, we can reduce the size of the solution space. We specifically focus on group-structured symmetries (invertible transformations). Additionally, we introduce an easy method for constructing equivariant network layers numerically, so the system designer need not solve the constraints by hand, as is typically done. We construct MDP homomorphic MLPs and CNNs that are equivariant under either a group of reflections or rotations. We show that such networks converge faster than unstructured baselines on CartPole, a grid world and Pong.

Reinforcement learning for recommendation (RL4Rec) methods are increasingly receiving attention as an effective way to improve long-term user engagement. However, applying RL4Rec online comes with risks: exploration may lead to periods of detrimental user experience. Moreover, few researchers have access to real-world recommender systems. Simulations have been put forward as a solution where user feedback is simulated based on logged historical user data, thus enabling optimization and evaluation without being run online. While simulators do not risk the user experience and are widely accessible, we identify an important limitation of existing simulation methods. They ignore the interaction biases present in logged user data, and consequently, these biases affect the resulting simulation. As a solution to this issue, we introduce a debiasing step in the simulation pipeline, which corrects for the biases present in the logged data before it is used to simulate user behavior. To evaluate the effects of bias on RL4Rec simulations, we propose a novel evaluation approach for simulators that considers the performance of policies optimized with the simulator. Our results reveal that the biases from logged data negatively impact the resulting policies, unless corrected for with our debiasing method. While our debiasing methods can be applied to any simulator, we make our complete pipeline publicly available as the Simulator for OFfline leArning and evaluation (SOFA): the first simulator that accounts for interaction biases prior to optimization and evaluation.

We develop ancestral Gumbel-Top-$k$ sampling: a generic and efficient method for sampling without replacement from discrete-valued Bayesian networks, which includes multivariate discrete distributions, Markov chains and sequence models. The method uses an extension of the Gumbel-Max trick to sample without replacement by finding the top $k$ of perturbed log-probabilities among all possible configurations of a Bayesian network. Despite the exponentially large domain, the algorithm has a complexity linear in the number of variables and sample size $k$. Our algorithm allows to set the number of parallel processors $m$, to trade off the number of iterations versus the total cost (iterations times $m$) of running the algorithm. For $m = 1$ the algorithm has minimum total cost, whereas for $m = k$ the number of iterations is minimized, and the resulting algorithm is known as Stochastic Beam Search. We provide extensions of the algorithm and discuss a number of related algorithms. We analyze the properties of ancestral Gumbel-Top-$k$ sampling and compare against alternatives on randomly generated Bayesian networks with different levels of connectivity. In the context of (deep) sequence models, we show its use as a method to generate diverse but high-quality translations and statistical estimates of translation quality and entropy.

The well-known Gumbel-Max trick for sampling from a categorical distribution can be extended to sample $k$ elements without replacement. We show how to implicitly apply this ’Gumbel-Top-$k$’ trick on a factorized distribution over sequences, allowing to draw exact samples without replacement using a Stochastic Beam Search. Even for exponentially large domains, the number of model evaluations grows only linear in $k$ and the maximum sampled sequence length. The algorithm creates a theoretical connection between sampling and (deterministic) beam search and can be used as a principled intermediate alternative. In a translation task, the proposed method compares favourably against alternatives to obtain diverse yet good quality translations. We show that sequences sampled without replacement can be used to construct low-variance estimators for expected sentence-level BLEU score and model entropy.

Diversity of environments is a key challenge that causes learned robotic controllers to fail due to the discrepancies between the training and evaluation conditions. Training from demonstrations in various conditions can mitigate - but not completely prevent - such failures. Learned controllers such as neural networks typically do not have a notion of uncertainty that allows to diagnose an offset between training and testing conditions, and potentially intervene. In this work, we propose to use Bayesian Neural Networks, which have such a notion of uncertainty. We show that uncertainty can be leveraged to consistently detect situations in high-dimensional simulated and real robotic domains in which the performance of the learned controller would be sub-par. Also, we show that such an uncertainty based solution allows making an informed decision about when to invoke a fallback strategy. One fallback strategy is to request more data. We empirically show that providing data only when requested results in increased data-efficiency.

Building models capable of generating structured output is a key challenge for AI and robotics. While generative models have been explored on many types of data, little work has been done on synthesizing lidar scans, which play a key role in robot mapping and localization. In this work, we show that one can adapt deep generative models for this task by unravelling lidar scans into a 2D point map. Our approach can generate high quality samples, while simultaneously learning a meaningful latent representation of the data. We demonstrate significant improvements against state-of-the-art point cloud generation methods. Furthermore, we propose a novel data representation that augments the 2D signal with absolute positional information. We show that this helps robustness to noisy and imputed input; the learned model can recover the underlying lidar scan from seemingly uninformative data.

In this work, we propose a novel method for training neural networks to perform singledocument extractive summarization without heuristically-generated extractive labels.We call our approach BANDITSUM as it treats extractive summarization as a contextual bandit (CB) problem, where the model receives a document to summarize (the context), and chooses a sequence of sentences to include in the summary (the action).A policy gradient reinforcement learning algorithm is used to train the model to select sequences of sentences that maximize ROUGE score.We perform a series of experiments demonstrating that BANDITSUM is able to achieve ROUGE scores that are better than or comparable to the state-of-the-art for extractive summarization, and converges using significantly fewer update steps than competing approaches.In addition, we show empirically that BANDIT-SUM performs significantly better than competing approaches when good summary sentences appear late in the source document.

In value-based reinforcement learning methods such as deep Q-learning, function approximation errors are known to lead to overestimated value estimates and suboptimal policies. We show that this problem persists in an actor-critic setting and propose novel mechanisms to minimize its effects on both the actor and the critic. Our algorithm builds on Double Q-learning, by taking the minimum value between a pair of critics to limit overestimation. We draw the connection between target networks and overestimation bias, and suggest delaying policy updates to reduce per-update error and further improve performance. We evaluate our method on the suite of OpenAI gym tasks, outperforming the state of the art in every environment tested.

In the pursuit of increasingly intelligent learning systems, abstraction plays a vital role in enabling sophisticated decisions to be made in complex environments. The options framework provides formalism for such abstraction over sequences of decisions. However most models require that options be given a priori, presumably specified by hand, which is neither efficient, nor scalable. Indeed, it is preferable to learn options directly from interaction with the environment. Despite several efforts, this remains a difficult problem. In this work we develop a novel policy gradient method for the automatic learning of policies with options. This algorithm uses inference methods to simultaneously improve all of the options available to an agent, and thus can be employed in an off-policy manner, without observing option labels. The differentiable inference procedure employed yields options that can be easily interpreted. Empirical results confirm these attributes, and indicate that our algorithm has an improved sample efficiency relative to state-of-the-art in learning options end-to-end.

Direct policy search has shown to be a successful method to optimize robot controller parameters. However, defining a good parametric form for the controller can be challenging for complex problems. Non-parametric methods provide a flexible alternative and are thus a promising tool in robot skill learning. In this paper, we investigate two nonparametric methods based on similar principles but utilizing differing computing schedules: an eager learner and a memory-based learner. We compare the methods experimentally on two different control problems. Furthermore, we define and evaluate a new `hybrid' controller that combines the strong points of both of these methods.

Direct contextual policy search methods learn to improve policy parameters and simultaneously generalize these parameters to different context or task variables. However, learning from high-dimensional context variables, such as camera images, is still a prominent problem in many real-world tasks. A naive application of unsupervised dimensionality reduction methods to the context variables, such as principal component analysis, is insufficient as task-relevant input may be ignored. In this paper, we propose a contextual policy search method in the model-based relative entropy stochastic search framework with integrated dimensionality reduction. We learn a model of the reward that is locally quadratic in both the policy parameters and the context variables. Furthermore, we perform supervised linear dimensionality reduction on the context variables by nuclear norm regularization. The experimental results show that the proposed method outperforms naive dimensionality reduction via principal component analysis and a state-of-the-art contextual policy search method.

Learning complex control policies from non-linear and redundant sensory input is an important challenge for reinforcement learning algorithms. Non-parametric methods that approximate values functions or transition models can address this problem, by adapting to the complexity of the data set. Yet, many current non-parametric approaches rely on unstable greedy maximization of approximate value functions, which might lead to poor convergence or oscillations in the policy update. A more robust policy update can be obtained by limiting the information loss between successive state-action distributions. In this paper, we develop a policy search algorithm with policy updates that are both robust and non-parametric. Our method can learn non- parametric control policies for infinite horizon continuous Markov decision processes with non-linear and redundant sensory representations. We investigate how we can use approximations of the kernel function to reduce the time requirements of the demanding non-parametric computations. In our experiments, we show the strong performance of the proposed method, and how it can be approximated efficiently. Finally, we show that our algorithm can learn a real-robot under-powered swing-up task directly from image data.

For many tasks, tactile or visual feedback is helpful or even crucial. However, designing controllers that take such high-dimensional feedback into account is non-trivial. Therefore, robots should be able to learn tactile skills through trial and error by using reinforcement learning algorithms. The input domain for such tasks, however, might include strongly correlated or non-relevant dimensions, making it hard to specify a suitable metric on such domains. Auto-encoders specialize in finding compact representations, where defining such a metric is likely to be easier. Therefore, we propose a reinforcement learning algorithm that can learn non-linear policies in continuous state spaces, which leverages representations learned using auto-encoders. We first evaluate this method on a simulated toy-task with visual input. Then, we validate our approach on a real-robot tactile stabilization task.

Accurate object shape knowledge provides important information for performing stable grasping and dexterous manipulation. When modeling an object using tactile sensors, touching the object surface at a fixed grid of points can be sample inefficient. In this paper, we present an active touch strategy to efficiently reduce the surface geometry uncertainty by leveraging a probabilistic representation of object surface. In particular, we model the object surface using a Gaussian process and use the associated uncertainty information to efficiently determine the next point to explore. We validate the resulting method for tactile object surface modeling using a real robot to reconstruct multiple, complex object surfaces.

Learning complex control policies from high-dimensional sensory input is a challenge for reinforcement learning algorithms. Kernel methods that approximate values functions or transition models can address this problem. Yet, many current approaches rely on instable greedy maximization. In this paper, we develop a policy search algorithm that integrates robust policy updates and kernel embeddings. Our method can learn non-parametric control policies for infinite horizon continuous MDPs with high-dimensional sensory representations. We show that our method outperforms related approaches, and that our algorithm can learn an underpowered swing-up task task directly from high-dimensional image data.

Most manipulation tasks can be decomposed into a sequence of phases, where the robot's actions have different effects in each phase. The robot can perform actions to transition between phases and, thus, alter the effects of its actions, e.g. grasp an object in order to then lift it. The robot can thus reach a phase that affords the desired manipulation. In this paper, we present an approach for exploiting the phase structure of tasks in order to learn manipulation skills more efficiently. Starting with human demonstrations, the robot learns a probabilistic model of the phases and the phase transitions. The robot then employs model-based reinforcement learning to create a library of motor primitives for transitioning between phases. The learned motor primitives generalize to new situations and tasks. Given this library, the robot uses a value function approach to learn a high-level policy for sequencing the motor primitives. The proposed method was successfully evaluated on a real robot performing a bimanual grasping task.

During grasping and other in-hand manipulation tasks maintaining a stable grip on the object is crucial for the task's outcome. Inherently connected to grip stability is the concept of slip. Slip occurs when the contact between the fingertip and the object is partially lost, resulting in sudden undesired changes to the objects state. While several approaches for slip detection have been proposed in the literature, they frequently rely on previous knowledge of the manipulated object. This previous knowledge may be unavailable, seeing that robots operating in real-world scenarios often must interact with previously unseen objects. In our work we explore the generalization capabilities of well known supervised learning methods, using random forest classifiers to create generalizable slip predictors. We utilize these classifiers in the feedback loop of an object stabilization controller. We show that the controller can successfully stabilize previously unknown objects by predicting and counteracting slip events.

In many complex robot applications, such as grasping and manipulation, it is difficult to program desired task solutions beforehand, as robots are within an uncertain and dynamic environment. In such cases, learning tasks from experience can be a useful alternative. To obtain a sound learning and generalization performance, machine learning, especially, reinforcement learning, usually requires sufficient data. However, in cases where only little data is available for learning, due to system constraints and practical issues, reinforcement learning can act suboptimally. In this paper, we investigate how model-based reinforcement learning, in particular the probabilistic inference for learning control method (Pilco), can be tailored to cope with the case of sparse data to speed up learning. The basic idea is to include further prior knowledge into the learning process. As Pilco is built on the probabilistic Gaussian processes framework, additional system knowledge can be incorporated by defining appropriate prior distributions, e.g. a linear mean Gaussian prior. The resulting Pilco formulation remains in closed form and analytically tractable. The proposed approach is evaluated in simulation as well as on a physical robot, the Festo Robotino XT. For the robot evaluation, we employ the approach for learning an object pick-up task. The results show that by including prior knowledge, policy learning can be sped up in presence of sparse data.

Phase transitions in manipulation tasks often occur when contacts between objects are made or broken. A switch of the phase can result in the robot's actions suddenly influencing different aspects of its environment. Therefore, the boundaries between phases often correspond to constraints or subgoals of the manipulation task. In this paper, we investigate how the phases of manipulation tasks can be learned from data. The task is modeled as an autoregressive hidden Markov model, wherein the hidden phase transitions depend on the observed states. The model is learned from data using the expectation-maximization algorithm. We demonstrate the proposed method on both a pushing task and a pepper mill turning task. The proposed approach was compared to a standard autoregressive hidden Markov model. The experiments show that the learned models can accurately predict the transitions in phases during the manipulation tasks.

Creating robots that can act autonomously in dynamic unstructured environments requires dealing with novel objects. Thus, an offline learning phase is not sufficient for recognizing and manipulating such objects. Rather, an autonomous robot needs to acquire knowledge through its own interaction with its environment, without using heuristics encoding human insights about the domain. Interaction also allows information that is not present in static images of a scene to be elicited. Out of a potentially large set of possible interactions, a robot must select actions that are expected to have the most informative outcomes to learn efficiently. In the proposed bottom-up probabilistic approach, the robot achieves this goal by quantifying the expected informativeness of its own actions in information-theoretic terms. We use this approach to segment a scene into its constituent objects. We retain a probability distribution over segmentations. We show that this approach is robust in the presence of noise and uncertainty in real-world experiments. Evaluations show that the proposed information-theoretic approach allows a robot to efficiently determine the composite structure of its environment. We also show that our probabilistic model allows straightforward integration of multiple modalities, such as movement data and static scene features. Learned static scene features allow for experience from similar environments to speed up learning for new scenes.

Creating robots that can act autonomously in dynamic, unstructured environments is a major challenge. In such environments, learning to recognize and manipulate novel objects is an important capability. A truly autonomous robot acquires knowledge through interaction with its environment without using heuristics or prior information encoding human domain insights. Static images often provide insufficient information for inferring the relevant properties of the objects in a scene. Hence, a robot needs to explore these objects by interacting with them. However, there may be many exploratory actions possible, and a large portion of these actions may be non-informative. To learn quickly and efficiently, a robot must select actions that are expected to have the most informative outcomes. In the proposed bottom-up approach, the robot achieves this goal by quantifying the expected informativeness of its own actions. We use this approach to segment a scene into its constituent objects as a first step in learning the properties and affordances of objects. Evaluations showed that the proposed information-theoretic approach allows a robot to efficiently infer the composite structure of its environment.