The Romanian Crosswords Competition Problem is a challenging, NP-hard constraint optimization problem. The task is to fill a $13 \times 13$ grid with criss-crossing words and up to 26 black cells. Words from two lists, the thematic list and the regular list, can be utilized. Each thematic word adds a number of points equal to its length to the score and the objective is to maximize the score. A recent AI approach to the problem constructs seeds, partial solutions with a high density of points in a subarea of the grid. Seeds are then completed to full solutions via a stochastic search. We present an approach to significantly extend the size of a seed which makes completion via stochastic search more successful. Experiments demonstrate an improvement of the completed solutions over published state-of-the-art algorithms.

The Romanian Crosswords Competition Problem is a challenging, NP-hard constraint optimization problem where state-of-the-art Artificial Intelligence has been lagging behind top human performance. The task is to construct a high-score grid with words from a thematic list, that changes every year, and from a regular list. Each thematic word in the solution gives a number of score points equal to its length. A recent approach to the problem generated grids with scores in the range of top human performance for the first time. However, such scores were obtained for only three years. We present new results with experiments on a larger scale, including an increase from three to eleven years. We produce grids with scores in the top-12 human range in six out of eleven years and discuss steps towards outperforming top humans.

In constraint optimization many problem instances remain challenging to current technology. We focus on the Romanian Crosswords Competition Problem. It is a challenging, NP-hard constraint optimization problem where state-of-the-art AI has been lagging significantly behind top human performance. We present an approach that first builds a core, a portion of the problem that will have a high contribution to the objective function. A core is grown into a seed, a partial solution with a subset of variables defined and instantiated. Seeds are further extended into full solutions. Our approach takes as input the size of a rectangular core to consider, and the locations of zero or more black cells inside the core. The results advance state-of-the-art substantially. We report a boost in the scores obtained, bringing our top solutions in the vicinity of top human entries.

Optimizing a machine learning (ML) pipeline has been an important topic of AI and ML. Despite recent progress, pipeline optimization remains a challenging problem, due to potentially many combinations to consider as well as slow training and validation. We present the BLDS algorithm for optimized algorithm selection (ML operations) in a fixed ML pipeline structure. BLDS performs multi-fidelity optimization for selecting ML algorithms trained with smaller computational overhead, while controlling its pipeline search based on multi-armed bandit and limited discrepancy search. Our experiments on well-known classification benchmarks show that BLDS is superior to competing algorithms. We also combine BLDS with hyperparameter optimization, empirically showing the advantage of BLDS.

AutoML automatically selects, composes and parameterizes machine learning algorithms into a workflow or pipeline of operations that aims at maximizing performance on a given dataset. Although current methods for AutoML achieved impressive results they mostly concentrate on optimizing fixed linear workflows. In this paper, we take a different approach and focus on generating and optimizing pipelines of complex directed acyclic graph shapes. These complex pipeline structure may lead to discovering hidden features and thus boost performance considerably. We explore the power of heuristic search and context-free grammars to search and optimize these kinds of pipelines. Experiments on various benchmark datasets show that our approach is highly competitive and often outperforms existing AutoML systems.

Crossword puzzles are a challenging game of skill in which humans have been competing for decades. Recently a heuristic-search-based Artificial Intelligence (AI) solver for Romanian crossword puzzles achieved competition-level scores. In this work in progress we tackle procedural content generation of crossword puzzles. Using genetic algorithms we evolve crossword puzzle instances on which the AI solver can achieve a high score. Since the solver takes a substantial time to solve each instance we first train a deep neural network to predict the solution score the solver would achieve. We then run the evolution of crossword puzzles with the network as the fast proxy fitness function. We show that doing so is an effective way of procedurally generating numerous crossword puzzles.

Heuristic search remains a leading approach to difficult combinatorial optimization problems. Search algorithms can utilize pruning based on comparing a target score with an admissible (optimistic) estimate of the best score that can be achieved from a given state. If the former is larger they prune the state. However, when the target score is too high the search can fail by exhausting the space without finding a solution. In this paper we show that such failed searches can still be valuable. Specifically, best partial solutions encountered in such failed searches can often bear a high similarity to the corresponding part of a full high-quality or even optimal solution. Thus, a new search for a full solution, with a lower target score, can start with a best known partial solution, rather than starting from scratch. We demonstrate our ideas in a constraint optimization problem modelled on the Romanian Crosswords Competition, a challenging problem where humans perform much better than computers. Utilizing partial solutions produced by a failed search cuts down the running time of an existing state-of-the-art solver by orders of magnitude on competition-level crossword puzzle instances and allows to solve more instances.

Finding a shortest path in a graph is at the core of many combinatorial search problems. A closely related problem refers to counting the number of shortest paths between two nodes. Such problems are solvable in polynomial time in the size of the graph. However, more realistic problem formulations could additionally specify constraints to satisfy. We study the problem of counting the shortest paths that are vertex disjoint and can satisfy additional constraints. Specifically, we look at the problems of counting vertex-disjoint shortest paths in edge-colored graphs, counting vertex-disjoint shortest paths with directional constraints, and counting vertex-disjoint shortest paths between multiple source-target pairs. We give a detailed theoretical analysis, and show formally that all of these three counting problems are NP-complete in general.

Recommender systems are used with increasing frequency in a wide variety of domains ranging from e- commerce to tourism, healthcare and online learning. However, the interaction with these systems generally tends to be limited to shallow feedback, such as providing ratings or filtering. Allowing users to interact with the recommender systems in a conversational environment brings opportunities in which the preferences can effectively be elicited from the users while the users can feel more in control of the whole process. However, when the existing non-interactive recommender systems are considered, it may not be easy to build an interactive layer directly on top of them. This is because there is already a great deal of modelling and work invested in the underlying algorithm and the system itself. Enabling interaction could mean rebuilding the whole solution from scratch, as the current design may not be able to consume preferences and information learnt from the user interaction online. In this paper, we propose the Interactive Recommender Framework, which converts non-interactive recommender solutions to conversational recommenders. We demonstrate how Interactive Recommender Framework can successfully enable interactivity on top of non-interactive recommender systems by integrating it into two different recommender algorithms from literature, and validate our solution through offline simulation experiments and online user studies.

Computing cycle-free solutions in cyclic AND/OR search spaces is an important AI problem. Previous work on optimal depth-first search strongly assumes the use of consistent heuristics, the need to keep all examined states in a transposition table, and the existence of solutions. We give a new theoretical analysis under relaxed assumptions where previous results no longer hold. We then present a generic approachto proving unsolvability, and apply it to RBFAOO and BLDFS, two state-of-the-art algorithms. We demonstrate the performance in domain-independent nondeterministic planning

Career goals represent a special case for recommender systems and require considering both short and long term goals. Recommendations must represent a trade off between relevance to the user, achievability and aspirational goals to move the user forward in their career. Users may have different motivations and concerns when looking for a new long term goal, so involving the user in the recommender process becomes all the more important than in other domains. Additionally, the cost to the user of making a bad decision is much higher than investing two hours in watching a movie they don't like or listening to an unappealing song. As a result, we feel career recommendations is a unique opportunity to truly engage the user in an interactive recommender as we believe they will invest the cognitive load. In this paper, we present an interactive career goal recommender framework that leverages the power of dialogue to allow the user interactively improve the recommendations and bring their own preferences to the system. The underlying recommendation algorithm is a novel solution that suggests both short and long term goals through utilizing the sequential patterns extracted from career trajectories that are enhanced with features of the supporting user profiles. The effectiveness of the proposed solution is demonstrated with extensive experiments on two real world data sets.

Search techniques, such as Monte Carlo Tree Search (MCTS) and Proof-Number Search (PNS), are effective in playing and solving games. However, the understanding of their performance in industrial applications is still limited. We investigate MCTS and Depth-First Proof-Number (DFPN) Search, a PNS variant, in the domain of Retrosynthetic Analysis (RA). We find that DFPN's strengths, that justify its success in games, have limited value in RA, and that an enhanced MCTS variant by Segler et al. significantly outperforms DFPN. We address this disadvantage of DFPN in RA with a novel approach to combine DFPN with Heuristic Edge Initialization. Our new search algorithm DFPN-E outperforms the enhanced MCTS in search time by a factor of 3 on average, with comparable success rates.

Recent research focuses beyond recommendation accuracy, towards human factors that influence the acceptance of recommendations, such as user satisfaction, trust, transparency and sense of control. We present a generic interactive recommender framework that can add interaction functionalities to non-interactive recommender systems. We take advantage of dialogue systems to interact with the user and we design a middleware layer to provide the interaction functions, such as providing explanations for the recommendations, managing users' preferences learnt from dialogue, preference elicitation and refining recommendations based on learnt preferences.

Single-agent pathfinding on grid maps can exploit online compiled knowledge produced offline and saved as a Compressed Path Database (CPD). Such a knowledge is distilled by performing repeated searches in a graph, where each node corresponds to a distinct grid cell, typically by algorithms such as Dijkstra's. All-pairs shortest paths (APSPs) are computed and the first move along a shortest path is persistently stored in the CPD. This way, an optimal move can efficiently be retrieved for any pair of source and target cells that is considered while the agent is navigating. However, a CPD supports a static grid, that is, a grid where each cell is permanently either traversable or non-traversable. Our work instead assumes that the cells in the map can undergo dynamic changes. Reasoning about the altered map would require a new CPD. As creating it from scratch is computationally expensive, we present techniques to repair an existing CPD. We prove that using our technique leads to correct and optimal solutions. Experiments demonstrate the benefits of our approach. When a single obstacle of a given size is added or removed, the repair costs often are a small fraction of a recomputation from scratch.

Multi-modal journey planning, which allows multiple types of transport within a single trip, is becoming increasingly popular, due to a strong practical interest and an increasing availability of data. In real life, transport networks feature uncertainty. Yet, most approaches assume a deterministic environment, making plans more prone to failures such as missed connections and major delays in the arrival. This paper presents an approach to computing optimal contingent plans in multi-modal journey planning. The problem is modeled as a search in an and/or state space. We describe search enhancements used on top of the AO* algorithm. Enhancements include admissible heuristics, multiple types of pruning that preserve the completeness and the optimality, and a hybrid search approach with a deterministic and a nondeterministic search. We demonstrate an NP-hardness result, with the hardness stemming from the dynamically changing distributions of the travel time random variables. We perform a detailed empirical analysis on realistic transport networks from cities such as Montpellier, Rome and Dublin. The results demonstrate the effectiveness of our algorithmic contributions, and the benefits of contingent plans as compared to standard sequential plans, when the arrival and departure times of buses are characterized by uncertainty.

We argue that chemistry should be the next grand challenge for Artificial Intelligence. The AI research community and humanity would benefit tremendously from focusing AI research on chemistry on a regular basis, as a benchmark as well as a real-world application domain. To support our position, we review the importance of chemical compound discovery and synthesis planning and discuss the properties of search spaces in a chemistry problem. Knowledge acquired in domains such as two-player board games or single-player puzzles places the AI community in a good position to solve critical problems in the chemistry domain. Yet, we show that searching in chemistry problems poses significant additional challenges that will have to be addressed. Finally, we envision how several AI areas like Natural Language Processing, Machine Learning, planning and search, are relevant for chemistry.

We present and evaluate diBOX, an algorithm for multi-agent path finding on strongly biconnected directed graphs. diBOX runs in polynomial time, computes suboptimal solutions and is complete for instances on strongly biconnected digraphs with at least two unoccupied positions. A detailed empirical analysis shows a good scalability for diBOX.

Quantum circuit compilation (QCC) is an important problem in the emerging field of quantum computing. The problem has a strong relevance to combinatorial search, as solving approaches recently published include constraint programming and temporal planning. In this paper, we focus on a complexity analysis of quantum circuit compilation. We formulate a makespan optimization problem based on QCC, and prove that the problem is NP-complete. To the best of our knowledge, this is the first study on the theoretical complexity of QCC.

Much of the literature on suboptimal, polynomial-time algorithms for multi-agent path finding focuses on undirected graphs, where motion is permitted in both directions along a graph edge. Despite this, traveling on directed graphs is relevant in navigation domains, such as path finding in games, and asymmetric communication networks.We consider multi-agent path finding on strongly biconnected directed graphs. We show that all instances with at least two unoccupied positions have a solution, except for a particular, degenerate subclass where the graph has a cyclic shape. We present diBOX, an algorithm for multi-agent path finding on strongly biconnected directed graphs. diBOX runs in polynomial time, computes suboptimal solutions and is complete for instances on strongly biconnected digraphs with at least two unoccupied positions. We theoretically analyze properties of the algorithm and properties of strongly biconnected directed graphs that are relevant to our approach. We perform a detailed empirical analysis of diBOX, showing a good scalability. To our knowledge, our work is the first study of multi-agent path finding focused on directed graphs.

Optimal heuristic search has been successful in many domains, including journey planning, route planning and puzzle solving. Existing work typically assumes that the cost of each action can easily be obtained. However, in many problems, the exact edge cost is expensive to compute. Existing search algorithms face a significant performance bottleneck, due to an excessive overhead associated with dynamically calculating exact edge costs. We present DEA*, an algorithm for problems with expensive edge cost computations. DEA* combines heuristic edge cost evaluations with delayed node expansions, reducing the number of exact edge computations. We formally prove that DEA* is optimal and it is efficient with respect to the number of exact edge cost computations. We empirically evaluate DEA* on multiple-worker routing problems where the exact edge cost is calculated by invoking an external multi-modal journey planning engine. The results demonstrate the effectiveness of our ideas in reducing the computational time and improving the solving ability. In addition, we show the advantages of DEA* in domain-independent planning, where we simulate that accurate edge costs are expensive to compute.

Real-time search algorithms are relevant to time-sensitive decision-making domains such as video games and robotics. In such settings, the agent is required to decide on each action under a constant time bound, regardless of the search space size. Despite recent progress, poor-quality solutions can be produced mainly due to state re-visitation. Different techniques have been developed to reduce such a re-visitation with state pruning showing promise. In this paper, we propose a novel pruning approach applicable to the wide class of real-time search algorithms. Given a local search space of arbitrary size, our technique aggressively prunes away all states in its interior, possibly adding new edges to maintain the connectivity of the search space frontier. An experimental evaluation shows that our pruning often improves the performance of a base real-time search algorithm by over an order of magnitude. This allows our implemented system to outperform state-of-the-art real-time search algorithms used in the evaluation.

Chasing multiple mobile targets with multiple agents is important in several applications, such as computer games and police chasing scenarios. Existing approaches can compute optimal policies. However, they have a limited scalability, as they implement expensive minimax searches. We introduce a sub-optimal but scalable approach that assigns individual agents to individual targets and that can dynamically re-compute such assignments. We provide a theoretical analysis, including upper bounds on the number of time steps required to solve an instance. In a detailed empirical evaluation on grid maps, our algorithm scales up very convincingly beyond the limits of previous methods. On small problems, where a comparison to a minimax approach is possible, the results demonstrate a good solution quality for our method.

This paper presents parallel dovetailing in a distributed-memory environment for exact MAP inference in graphical models. Parallel dovetailing is a simple procedure which performs multiple searches in parallel with different parameter configurations. We evaluate empirically the performance of parallel dovetailing with three state-of-the-art AND/OR search algorithms in solving various MAP inference benchmarks. Our results clearly show that parallel dovetailing is effective, yielding considerable speedups and improving the solving abilities of these state-of-the-art baseline methods.

Optimal multi-modal journey planning under uncertainty is a challenging problem, due in part to an increased branching factor generated by nondeterministic actions. Deterministic search, which ignores all uncertainty, can be much faster, but deterministic plans lack correctness and optimality guarantees in the uncertainty-aware domain.

Much of the literature on multi-agent path finding focuses on undirected graphs, where motion is permitted in both directions along a graph edge. Despite this, travelling on directed graphs is relevant in navigation domains, such as pathfinding in games, and asymmetric communication networks. We consider multi-agent path finding on strongly biconnected directed graphs. We show that all instances with at least two unoccupied positions can be solved or proven unsolvable. We present a polynomial-time algorithm for this class of problems, and analyze its complexity. Our work may be the first formal study of multi-agent path finding on directed graphs.

In this work we give a first tractability analysis of Compressed Path Databases, space efficient oracles used to very quickly identify the first arc on a shortest path. We study the complexity of computing an optimal compressed path database for general directed and undirected graphs. We find that in both cases the problem is NP-complete. We also show that, for graphs which can be decomposed along articulalion points, the problem can be decomposed into independent parts, with a corresponding reduction in its level of difficulty. In particular, this leads to simple and tractable algorithms which yield optimal compression results for trees.

The paper presents and evaluates the power of parallel search for exact MAP inference in graphical models. We introduce a new parallel shared-memory recursive best-first AND/OR search algorithm, called SPRBFAOO, that explores the search space in a best-first manner while operating with restricted memory. Our experiments show that SPRBFAOO is often superior to the current state-of-the-art sequential AND/OR search approaches, leading to considerable speed-ups (up to 7-fold with 12 threads), especially on hard problem instances.

The Grid-Based Path Planning Competition has just completed its third iteration. The entriesused in the competition have improved significantly during this time, changing the view ofthe state of the art of grid-based pathfinding. Furthermore, the entries from the competition have beenmade publicly available, improving the ability of researchers to compare their work. Thispaper summarizes the entries to the 2014 competition, presents the 2014 competition results,and talks about what has been learned and where there is room for improvement.

Knowledge on the Web relies heavily on multi-relational representations, such as RDF and Schema.org. Automatically extracting knowledge from documents and linking existing databases are common approaches to construct multi-relational data. Complementary to such approaches, there is still a strong demand for manually encoding human expert knowledge. For example, human annotation is necessary for constructing a common-sense knowledge base, which stores facts implicitly shared in a community, because such knowledge rarely appears in documents. As human annotation is both tedious and costly, an important research challenge is how to best use limited human resources, whiles maximizing the quality of the resulting dataset. In this paper, we formalize the problem of dataset construction as active learning problems and present the Active Multi-relational Data Construction (AMDC) method. AMDC repeatedly interleaves multi-relational learning and expert input acquisition, allowing us to acquire helpful labels for data construction. Experiments on real datasets demonstrate that our solution increases the number of positive triples by a factor of 2.28 to 17.0, and that the predictive performance of the multi-relational model in AMDC achieves the highest or comparable to the best performance throughout the data construction process.

We introduce a novel approach to Compressed Path Databases, space efficient oracles used to very quickly identify the first edge on a shortest path. Our algorithm achieves query running times on the 100 nanosecond scale, being significantly faster than state-of-the-art first-move oracles from the literature. Space consumption is competitive, due to a compression approach that rearranges rows and columns in a first-move matrix and then performs run length encoding (RLE) on the contents of the matrix. One variant of our implemented system was, by a convincing margin, the fastest entry in the 2014 Grid-Based Path Planning Competition. We give a first tractability analysis for the compression scheme used by our algorithm. We study the complexity of computing a database of minimum size for general directed and undirected graphs. We find that in both cases the problem is NP-complete. We also show that, for graphs which can be decomposed along articulation points, the problem can be decomposed into independent parts, with a corresponding reduction in its level of difficulty. In particular, this leads to simple and tractable algorithms with linear running time which yield optimal compression results for trees.

In moving target search, the objective is to guide a hunter agent to catch a moving prey. Even though in game applications maps are always available at developing time, current approaches to moving target search do not exploit preprocessing to improve search performance. In this paper, we propose MtsCopa, an algorithm that exploits precomputed information in the form of compressed path databases (CPDs), and that is able to guide a hunter agent in both known and partially known terrain. CPDs have previously been used in standard, fixed-target pathfinding but had not been used in the context of moving target search. We evaluated MtsCopa over standard game maps. Our speed results are orders of magnitude better than current state of the art. The time per individual move is improved, which is important in real-time search scenarios, where the time available to make a move is limited. Compared to state of the art, the number of hunter moves is often better and otherwise comparable, since CPDs provide optimal moves along shortest paths. Compared to previous successful methods, such as I-ARA*, our method is simple to understand and implement. In addition, we prove MtsCopa always guides the agent to catch the prey when possible.

Recommender systems can be employed to assist users in complex decision making processes. This paper presents a multi-criteria housing recommender system which takes into account not just features of a home, such as rent, but also the transportation links to user specified locations. First, we describe an efficient multi-hop journey time calculator. Second, we introduce a mechanism to find the optimal solutions for multi-criteria evaluation, where a balanced trade-off between the target goals is found. Finally, we present a user study to demonstrate the potential of such a system.

We introduce a novel preprocessing-based algorithm to solve the problem of determining the first arc of a shortest path in sparse graphs. Our algorithm achieves query running times on the 100 nanosecond scale, being significantly faster than state-of-the-art first-move oracles from the literature. Space consumption is competitive, due to a compression approach that rearranges rows and columns in a first-move matrix and then performs run length encoding (RLE) on the contents of the matrix.

The increasing availability of “utility computing” resources such as clouds, grids, and massively parallel shared clusters can provide practically unlimited processing and memory capacity on demand, at some cost per unit of resource usage. This requires a new perspective in the design and evaluation of parallel search algorithms. Previous work in parallel search implicitly assumed ownership of a cluster with a static amount of CPU cores and RAM, and emphasized wallclock runtime. With utility computing resources, trade-offs between performance and monetary costs must be considered. This paper considers dynamically increasing the usage of utility computing resources until a problem is solved. Efficient resource allocation policies are analyzed in comparison with an optimal allocation strategy. We evaluate our iterative allocation strategy by applying it to the HDA* parallel search algorithm. The experimental results validate our theoretical predictions. They show that, in practice, the costs incurred by iterative allocation are reasonably close to an optimal (but a priori unknown) policy, and are significantly better than the worst-case analytical bounds.

Most existing pathfinding methods are based on runtime search. Despite an impressive progress achieved in recent years, search-based methods could explore, in bad cases, large portions of a map, leading to an increased running time. We take a significantly different approach from search-based methods. Given a map, our program precomputes all-pairs shortest paths (APSP) information. APSP data are compressed, reducing the size dramatically with no information loss. We call the resulting data a compressed path database (CPD). At runtime, optimal moves are retrieved one by one until a complete (or partial, if desired) optimal path is retrieved. CPDs lead to a fast path computation, eliminating the need for runtime search. The first move lag is very low, as each move is retrieved independently from subsequent moves. The price to pay includes a significant preprocessing time and memory to build and store a CPD.

The increasing availability of “utility computing” resources such as clouds, grids, and massively parallel shared clusters can provide practically unlimited processing and memory capacity on demand, at some cost per unit of resource usage. This requires a new perspective in the design and evaluation of parallel search algorithms. Previous work in parallel search implicitly assumed ownership of a cluster with a static amount of CPU cores and RAM, and emphasized wallclock runtime. With utility computing resources, trade-offs between performance and monetary costs must be considered. This paper considers dynamically increasing the usage of utility computing resources until a problem is solved. Efficient resource allocation policies are analyzed in comparison with an optimal allocation strategy. We evaluate our iterative allocation strategy by applying it to the HDA* parallel search algorithm. The experimental results validate our theoretical predictions. They show that, in practice, the costs incurred by iterative allocation are reasonably close to an optimal (but a priori unknown) policy, and are significantly better than the worst-case analytical bounds.

MAPP has been previously shown as a state-of-the-art multi-agent path planning algorithm on criteria including scalability and success ratio (i.e., percentage of solved units) on realistic game maps. MAPP further provides a formal characterization of problems it can solve, and low-polynomial upper bounds on the resources required. However, until now, MAPP's solution quality had not been extensively analyzed. In this work we empirically analyze the quality of MAPP's solutions, using multiple quality criteria such as the total travel distance, the makespan and the sum of actions (including move and wait actions). We also introduce enhancements that improve MAPP's solution quality significantly. For example, the sum of actions is cut to half on average. The improved MAPP is competitive in terms of solution quality with FAR and WHCA*, two successful algorithms from the literature, and maintains its advantages on different performance criteria, such as scalability, success ratio, and ability to tell apriori if it will succeed in the instance at hand. As optimal algorithms have limited scalability, evaluating the quality of the solutions provided by suboptimal algorithms is another important topic. Using lower bounds of optimal values, we show that MAPP's solutions have a reasonable quality. For example, MAPP's total travel distance is on average 19% longer than a lower bound on the optimal value.

Scaling up the number of simultaneously moving units in pathfinding problems to hundreds, or even thousands, is well beyond the capability of theoretically optimal algorithms in practice, which is consistent with existing intractability results. Significant scalability can be achieved by trading off solution optimality, which motivates evaluating the quality of suboptimal solutions, especially in instances much larger than can be handled by optimal algorithms. We consider pathfinding in uniform cost grid maps, and we study the solution quality using the three most common quality criteria, total travel distance, sum of actions, and makespan. We focus on MAPP, which has been shown as state-of-the-art in terms of scalability and success ratio (i.e., percentage of solved units) on realistic game grid maps. Until now, the quality of MAPP's solutions had not been as extensively analyzed. We introduced enhancements that significantly improve MAPP's solution quality. For example, its sum of actions is cut to half on average. MAPP becomes competitive in terms of solution quality with FAR and WHCA*, two successful algorithms from the literature. To evaluate the quality of suboptimal solutions in instances beyond the capability of optimal algorithms, we use lower bounds of optimal values to show our solutions have a reasonable quality. For example, MAPP's average total travel distance is 19 percent longer than the lower bound.

Navigating multiple mobile units on grid maps is an NP-complete problem [4, 1] with many real-life applications. Centralized search in the combined state space of all units scales very poorly. Previous approaches that decompose the initial problem into a series of smaller searches, such as FAR [5] and WHCA* [2], can significantly improve scalability and speed. However, such methods are incomplete. They provide no guarantees with respect to the total running time, and are unable to apriori tell whether they would succeed in finding a solution to a given instance. More recent algorithms, such as MAPP [6] and BIBOX [3], are complete on well-specified subclasses of problems. They also provide low-polynomial upper bounds for the running time, the solution length and the memory requirements. However, their empirical speed and scalability, compared to incomplete methods, has been an open question. In this paper we take steps towards bridging the gap between the two categories of algorithms, combining strengths specific to each of them. We extended MAPP to improve its completeness range, solution quality, and total runtime, addressing main bottlenecks of the original algorithm. We performed the first empirical analysis of MAPP, showing that the enhanced MAPP has better success ratio and scalability than state-of-the-art incomplete algorithms, and is competitive in running times, with at most 92% longer solutions, while maintaining the theoretical properties of the original MAPP.

HDA* is a simple, parallelization of A* where work is asynchronously distributed among the nodes by a global hash function. Using up to 1024 cores on a large distributed memory cluster, we evaluate HDA* for a domain-independent planner as well an application-specific 24-puzzle solver. We show that HDA* scales fairly well on a large cluster using up to 1024 cores. Our analysis of the scaling behavior shows that on a cluster of multicore nodes, using only a subset of the available cores and leaving some cores idle can, surprisingly, lead to better results.

Recent contributions to advancing planning from the classical model to more realistic problems include using temporal logic such as LTL to express desired properties of a solution plan. This paper introduces a planning model that combines temporally extended goals and uncontrollable events. The planning task is to reach a state such that all event sequences generated from that state satisfy the problem's temporally extended goal. A real-life application that motivates this work is to use planning to configure a system in such a way that its subsequent, non-deterministic internal evolution (nominal behavior) is guaranteed to satisfy a condition expressed in temporal logic.

Despite recent progress in AI planning, many benchmarks remain challenging for current planners. In many domains, the performance of a planner can greatly be improved by discovering and exploiting information about the domain structure that is not explicitly encoded in the initial PDDL formulation. In this paper we present and compare two automated methods that learn relevant information from previous experience in a domain and use it to solve new problem instances. Our methods share a common four-step strategy. First, a domain is analyzed and structural information is extracted, then macro-operators are generated based on the previously discovered structure. A filtering and ranking procedure selects the most useful macro-operators. Finally, the selected macros are used to speed up future searches. We have successfully used such an approach in the fourth international planning competition IPC-4. Our system, Macro-FF, extends Hoffmann's state-of-the-art planner FF 2.3 with support for two kinds of macro-operators, and with engineering enhancements. We demonstrate the effectiveness of our ideas on benchmarks from international planning competitions. Our results indicate a large reduction in search effort in those complex domains where structural information can be inferred.