To achieve high performance, autonomous systems, such as science explorers, should adapt to the environment to improve utility gained, as well as robustness. Flexibility during temporal plan execution has been explored extensively to improve robustness, where flexibility exists both in activity choices and schedules. These problems are framed as conditional constraint networks over temporal constraints. However, flexibility has been exploited in a limited form to improve utility. Prior work considers utility in choice or schedule, but not their coupling. To exploit fully flexibility, we introduce conditional simple temporal networks with preference (CSTNP), where preference is a function over both choice and schedule. Enumerating best solutions to a CSTNP is challenging due to the cost of scheduling a candidate STPP and the exponential number of candidates. Our contribution is an algorithm for enumerating solutions to CSTNPs efficiently, called A star with bounding conflicts (A*BC), and a novel variant of conflicts, called bounding conflicts, for learning heuristic functions. A*BC interleaves Generate, Test, and Bound. When A*BC bounds a candidate, by solving a STPP, it generates a bounding conflict, denoting neighboring candidates with similar bounds. A*BC's generator then uses these conflicts to steer away from sub-optimal candidates.

Optimization-based heuristics may offer very good estimates. But, calculatingthem may be time consuming, especially if the optimization problem isintractable. This raises the question of their applicability. This papersummarizes early work from the year 2000 on optimization-based heuristics inthe context of PDBs for the Tile-Puzzle. We show that an admissible heuristicbased on Vertex-Cover (VC) can be calculated in reasonable time over a largecollection of small PDBs. When larger PDBs are involved we suggest the idea ofusing another lookup table that precalculates and stores all possible relevantVC values. This table can be later looked up in a constant time during thesearch. We discuss the conditions under which this idea can be generalized.Experimental results demonstrate the applicability of these two ideas on the15- and 24-Puzzle. The first idea appeared in (Felner, Korf and Hanan, 2004) but the secondidea is presented here for the first time.

The Atari 2600 games supported in the Arcade Learning Environment (Bellemare et al. 2013) all feature aknown initial (RAM) state and actions that have deterministic effects. Classical planners, however, cannot be used for selecting actions for two reasons: first, nocompact PDDL-model of the games is given, and more importantly, the action effects and goals are not known a priori. Moreover, in these games there is usually no set of goals to be achieved but rewards to be collected. These features do not preclude the use of classical algorithms like breadth-first search or Dijkstra’s algorithm, but these methods are not effective over large state spaces. We thus turn to a different class of classical planning algorithms introduced recently that perform a structured exploration of the state space; namely, like breadth-first search and Dijkstra’s algorithm they are“blind” and hence do not require prior knowledge of state transitions, costs (rewards) or goals, and yet, like heuristic search algorithms, they have been shown to be effective for solving problems over huge state spaces.The simplest such algorithm, called Iterated Width or IW, consists of a sequence of calls IW(1), IW(2), . . . ,IW(k) where IW(i) is a breadth-first search in which a state is pruned when it is not the first state in the search to make true some subset of i atoms. The empirical results over 54 games suggest that the performance of IW with the k parameter fixed to 1, i.e., IW(1), is at the level of the state of the art represented by UCT. A simple best-first variation of IW that combines exploration and exploitation proves to be very competitive as well.

Answering point-to-point distance queries is important inmany applications, including games, robotics and vehiclerouting in operations research. Searching in a graph to answer distance queries on demandcan often be too slow.An alternative strategy, taken in methods such asTransit and Hub Labels, is to pre-compute information that can help computedistances much faster.To be practical, such methods need to generate muchless preprocessed data than a naive all-pairs distance table. We present Heuristic-Aid Compressed Distance Databases (HCDs),pre-computed data structures based on the observation thatheuristic distance estimations can sometimes coincide with true distances.Compared to a naive all-pairs distance table,we report compression factors of two to three orders of magnitude in a wide range ofmaps, reducing the memory usage to a reasonable size. Comparedto compressed path databases, our approachgenerally generates smaller databases, and answers query distances faster.

The paper explores the potential of weighted best-first search schemes as anytime optimization algorithms for solving graphical models tasks such as MPE (Most Probable Explanation) or MAP (Maximum a Posteriori) and WCSP (Weighted Constraint Satisfaction Problem). While such schemes were widely investigated for path-finding tasks, their application for graphical models was largely ignored, possibly due to their memory requirements. Compared to the depth-first branch and bound, which has long been the algorithm of choice for optimization in graphical models, a valuable virtue of weighted best-first search is that they are w-optimal, i.e. when terminated, they return a solution cost C and a weight w, such that C < = wC*, where C* is the optimal cost. We report on a significant empirical evaluation, demonstrating the usefulness of weighted best-first search as approximation anytime schemes (that have suboptimality bounds) and compare against one of the best depth-first branch and bound solver to date. We also investigate the impact of different heuristic functions on the behaviour of the algorithms.

Stochastic Physical Search (SPS) refers to the search for an item in a physical environment where the item's price is stochastic, and where the cost to obtain the item includes both travel and purchase costs. This type of problem models task planning scenarios where the cost of completing an objective at a location is drawn from a probability distribution, reflecting the influence of unknown factors. Prior work on this domain has focused on solutions where the expected cost is minimized. Recently, SPS problems with other objectives have been proposed and theoretically analyzed, in particular when either the budget or the desired probability of success is fixed. However, general optimal solvers for these new variants do not yet exist. We present algorithms for optimal solution of these variants on general graphs. We formulate them as mixed integer linear programming problems, and solve them using an off-the-shelf MILP solver. We then develop custom branch and bound algorithms which result in a dramatic reduction in computation speed. Using these algorithms, we generate empirical insights into the hardness landscape of the fixed budget and fixed probability of success SPS variants.

We examine a decomposition approach to find good quality feasible solutions. In particular, we studya method to reduce the search-space by decomposing a problem into two partitions, where the second partition (i.e., the subproblem) contains the fixed solution of the first (i.e., the master). This type of approach is usually motivated by the presence of two sub-problems that are each more easily solved by different methods. Our work is motivated by methods for which it is nontrivial to return a strong `no-good', `Benders feasibility', or 'optimality' cut. Instead, we focus our attention on a uni-directional decomposition approach. Instead of providing a relaxation of the sub-problem for the master problem, as in Benders decomposition, we provide an approximation of the sub-problem. Thus, we aim at finding good quality feasible solutions in the first iteration. While the quality of the approximation itself affects the impact of this approach, we illustrate that even using a simple approximation can havestrong positive impact on two examples: the Travelling Purchaser Problem and a Mine Planning Problem.

In this paper we compare the abstraction methods used for state space search and planning in Artificial Intelligence with the state space relaxation methods used in Operations Research for various optimization problems such as the Travelling Salesman problem (TSP). Although developed independently, these methods are based on exactly the same general idea: lower bounds on distances in a given state space can be derived by computing exact distances in a ``simplified" state space. Our aim is to describe these methods so that the two communities understand what each other has done and can begin to work together.

Algorithm portfolios seek to determine an effective set of algorithms that can be used within an algorithm selection framework to solve problems. A limited number of these portfolio studies focus on generating different versions of a target algorithm using different parameter configurations. In this paper, we employ a Design of Experiments (DOE) approach to determine a promising range of values for each parameter of an algorithm. These ranges are further processed to determine a portfolio of parameter configurations, which would be used within two online Algorithm Selection approaches for solving different instances of a given combinatorial optimization problem effectively. We apply our approach on a Simulated Annealing-Tabu Search (SA-TS) hybrid algorithm for solving the Quadratic Assignment Problem (QAP) as well as an Iterated Local Search (ILS) on the Travelling Salesman Problem (TSP). We also generate a portfolio of parameter configurations using best-of-breed parameter tuning approaches directly for the comparison purpose. Experimental results show that our approach lead to improvements over best-of-breed parameter tuning approaches.

Learning to provide appropriate assistance to people indifferent situations is an extremely important, but insufficientlyinvestigated machine learning task. Applications includehuman-robot and human-computer interactions settings to maximizing the benefits of assistive technologies. Three key challenges must be overcome to appropriately address this task: Complexity: the space of possible assistive policies can be very large, making many existing methods (e.g., fromreinforcement learning) too data inefficient to be practical. Noise and misspecification: observed human behavior is often noisy and parametric formulations that reduce complexity will typically suffer from model misspecification,leading to unboundedly sub-optimal assistance. Biasedness: data available for learning a model is biased by previously provided assistive actions, violating the typical assumptions of supervised learning. We develop a general framework for learning to assist in single intervention settings. The framework narrows the search for effective assistance by viewing previous behavior under assistance through a restricted set of statistics. Assistive policies for the worst-case context-assistance-outcome relationships satisfying these statistics are obtained. We embed the problem of learning how to assist users in cursor based target pointing tasks into this framework and outline its usage.