Ariel Felner

Theta*: any-angle path planning on grids

Grids with blocked and unblocked cells are often used to represent terrain in computer games and robotics. However, paths formed by grid edges can be sub-optimal and unrealistic looking, since the possible headings are artificially constrained. We present Theta*, a variant of A*, that propagates informati on along grid edges without constraining the paths to grid edges. Theta* is simple, fast and finds short and realistic looking paths. We compare Theta* against both Field D*, the only other variant of A* that propagates information along grid edges without constraining the paths to grid edges, and A* with post-smoothed paths. Although neither path planning method is guaranteed to find shortest paths, we show experimentally that Theta* finds shorter and more realistic looking paths than either of these existing techniques.

Disjoint pattern database heuristics

We describe a new technique for designing more accurate admissible heuristic evaluation functions, based on pattern databases [J. Culberson, J. Schaeffer, Comput. Intelligence 14 (3) (1998) 318-334]. While many heuristics, such as Manhattan distance, compute the cost of solving individual subgoals independently, pattern databases consider the cost of solving multiple subgoals simultaneously. Existing work on pattern databases allows combining values from different pattern databases by taking their maximum. If the subgoals can be divided into disjoint subsets so that each operator only affects subgoals in one subset, then we can add the pattern-database values for each subset, resulting in a more accurate admissible heuristic function. We used this technique to improve performance on the Fifteen Puzzle by a factor of over 2000, and to find optimal solutions to 50 random instances of the Twenty-Four Puzzle.

BnB-ADOPT: an asynchronous branch-and-bound DCOP algorithm

Distributed constraint optimization (DCOP) problems are a popular way of formulating and solving agent-coordination problems. It is often desirable to solve DCOP problems optimally with memory-bounded and asynchronous algorithms. We introduce Branch-and-Bound ADOPT (BnB-ADOPT), a memory-bounded asynchronous DCOP algorithm that uses the message passing and communication framework of ADOPT, a well known memory-bounded asynchronous DCOP algorithm, but changes the search strategy of ADOPT from best-first search to depth-first branch-and-bound search. Our experimental results show that BnB-ADOPT is up to one order of magnitude faster than ADOPT on a variety of large DCOP problems and faster than NCBB, a memory-bounded synchronous DCOP algorithm, on most of these DCOP problems.

Additive pattern database heuristics

We explore a method for computing admissible heuristic evaluation functions for search problems. It utilizes pattern databases (Culberson & Schaeffer, 1998), which are precomputed tables of the exact cost of solving various subproblems of an existing problem. Unlike standard pattern database heuristics, however, we partition our problems into disjoint sub-problems, so that the costs of solving the different subproblems can be added together without overestimating the cost of solving the original problem. Previously (Korf & Felner, 2002) we showed how to statically partition the sliding-tile puzzles into disjoint groups of tiles to compute an admissible heuristic, using the same partition for each state and problem instance. Here we extend the method and show that it applies to other domains as well. We also present another method for additive heuristics which we call dynamically partitioned pattern databases. Here we partition the problem into disjoint subproblems for each state of the search dynamically. We discuss the pros and cons of each of these methods and apply both methods to three different problem domains: the sliding-tile puzzles, the 4-peg Towers of Hanoi problem, and finding an optimal vertex cover of a graph. We find that in some problem domains, static partitioning is most effective. while in others dynamic partitioning is a better choice. In each of these problem domains, either statically partitioned or dynamically partitioned pattern database heuristics are the best known heuristics for the problem.

A general theory of additive state space abstractions

Informally, a set of abstractions of a state space S is additive if the distance between any two states in S is always greater than or equal to the sum of the corresponding distances in the abstract spaces. The first known additive abstractions, called disjoint pattern databases, were experimentally demonstrated to produce state of the art performance on certain state spaces. However, previous applications were restricted to state spaces with special properties, which precludes disjoint pattern databases from being defined for several commonly used testbeds, such as Rubiks Cube, TopSpin and the Pancake puzzle. In this paper we give a general definition of additive abstractions that can be applied to any state space and prove that heuristics based on additive abstractions are consistent as well as admissible. We use this new definition to create additive abstractions for these testbeds and show experimentally that well chosen additive abstractions can reduce search time substantially for the (18,4)-TopSpin puzzle and by three orders of magnitude over state of the art methods for the 17-Pancake puzzle. We also derive a way of testing if the heuristic value returned by additive abstractions is provably too low and show that the use of this test can reduce search time for the 15-puzzle and TopSpin by roughly a factor of two.

KBFS: K-Best-First Search

We introduce a new algorithm, K-best-first search (KBFS), which is a generalization of the well known best-first search (BFS). In KBFS, each iteration simultaneously expands the K best nodes from the open-list (rather than just the best as in BFS). We claim that KBFS outperforms BFS in domains where the heuristic function has large errors in estimation of the real distance to the goal state or does not predict dead-ends in the search tree. We present empirical results that confirm this claim and show that KBFS outperforms BFS by a factor of 15 on random trees with dead-ends, and by a factor of 2 and 7 on the Fifteen and Twenty-Four tile puzzles, respectively. KBFS also finds better solutions than BFS and hill-climbing for the number partitioning problem. KBFS is only appropriate for finding approximate solutions with inadmissible heuristic functions.

Predicting the performance of IDA* using conditional distributions

Korf, Reid, and Edelkamp introduced a formula to predict the number of nodes IDA* will expand on a single iteration for a given consistent heuristic, and experimentally demonstrated that it could make very accurate predictions. In this paper we show that, in addition to requiring the heuristic to be consistent, their formulas predictions are accurate only at levels of the brute-force search tree where the heuristic values obey the unconditional distribution that they defined and then used in their formula. We then propose a new formula that works well without these requirements, i.e., it can make accurate predictions of IDA*s performance for inconsistent heuristics and if the heuristic values in any level do not obey the unconditional distribution. In order to achieve this we introduce the conditional distribution of heuristic values which is a generalization of their unconditional heuristic distribution. We also provide extensions of our formula that handle individual start states and the augmentation of IDA* with bidirectional pathmax (BPMX), a tech nique for propagating heuristic values when inconsistent heuristics are used. Experimental results demonstrate the accuracy of our new method and all its variations.

Maximizing over multiple pattern databases speeds up heuristic search

A pattern database (PDB) is a heuristic function stored as a lookup table. This paper considers how best to use a fixed amount (m units) of memory for storing pattern databases. In particular, we examine whether using n pattern databases of size m/n instead of one pattern database of size m improves search performance. In all the state spaces considered, the use of multiple smaller pattern databases reduces the number of nodes generated by IDA*. The paper provides an explanation for this phenomenon based on the distribution of heuristic values that occur during search.

Conflict-based search for optimal multi-agent pathfinding

In the multi agent path finding problem (MAPF) paths should be found for several agents, each with a different start and goal position such that agents do not collide. Previous optimal solvers applied global A*-based searches. We present a new search algorithm called Conflict Based Search (CBS). CBS is a two-level algorithm. At the high level, a search is performed on a tree based on conflicts between agents. At the low level, a search is performed only for a single agent at a time. In many cases this reformulation enables CBS to examine fewer states than A* while still maintaining optimality. We analyze CBS and show its benefits and drawbacks. Experimental results on various problems shows a speedup of up to a full order of magnitude over previous approaches.

Compressed pattern databases

A pattern database (PDB) is a heuristic function implemented as a lookup table that stores the lengths of optimal solutions for subproblem instances. Standard PDBs have a distinct entry in the table for each subproblem instance. In this paper we investigate compressing PDBs by merging several entries into one, thereby allowing the use of PDBs that exceed available memory in their uncompressed form. We introduce a number of methods for determining which entries to merge and discuss their relative merits. These vary from domain-independent approaches that allow any set of entries in the PDB to be merged, to more intelligent methods that take into account the structure of the problem. The choice of the best compression method is based on domain-dependent attributes. We present experimental results on a number of combinatorial problems, including the four-peg Towers of Hanoi problem, the sliding-tile puzzles, and the Top-Spin puzzle. For the Towers of Hanoi, we show that the search time can be reduced by up to three orders of magnitude by using compressed PDBs compared to uncompressed PDBs of the same size. More modest improvements were observed for the other domains.