Finding high quality plans for large planning problems is hard. Although some current anytime planners are often able to improve plans quickly, they tend to reach a limit at which the plans produced are still very far from the best possible, but these planners fail to find any further improvement, even when given several hours of runtime. We present an approach to continuing plan quality optimisation at larger time scales, and its implementation in a system called BDPO2. Key to this approach is a decomposition into subproblems of improving parts of the current best plan. The decomposition is based on block deordering, a form of plan deordering which identifies hierarchical plan structure. BDPO2 can be seen as an application of the large neighbourhood search (LNS) local search strategy to planning, where the neighbourhood of a plan is defined by replacing one or more subplans with improved subplans. On-line learning is also used to adapt the strategy for selecting subplans and subplanners over the course of plan optimisation. Even starting from the best plans found by other means, BDPO2 is able to continue improving plan quality, often producing better plans than other anytime planners when all are given enough runtime. The best results, however, are achieved by a combination of different techniques working together.

Bayesian Optimization (BO) is a data-efficient method for global black-box optimization of an expensive-to-evaluate fitness function. BO typically assumes that computation cost of BO is cheap, but experiments are time consuming or costly. In practice, this allows us to optimize ten or fewer critical parameters in up to 1,000 experiments. But experiments may be less expensive than BO methods assume: In some simulation models, we may be able to conduct multiple thousands of experiments in a few hours, and the computational burden of BO is no longer negligible compared to experimentation time. To address this challenge we introduce a new Dimension Scheduling Algorithm (DSA), which reduces the computational burden of BO for many experiments. The key idea is that DSA optimizes the fitness function only along a small set of dimensions at each iteration. This DSA strategy (1) reduces the necessary computation time, (2) finds good solutions faster than the traditional BO method, and (3) can be parallelized straightforwardly. We evaluate the DSA in the context of optimizing parameters of dynamic models of microalgae metabolism and show faster convergence than traditional BO.

Video-game designers often tessellate continuous 2-dimensional terrain into a grid of blocked and unblocked square cells. The three main ways to calculate short paths on such a grid are to determine truly shortest paths, shortest vertex paths and shortest grid paths, listed here in decreasing  order of computation time and increasing order of resulting path length. We show that, for both vertex and grid paths on both 4-neighbor and 8-neighbor grids, placing vertices at cell corners rather than at cell centers tends to result in shorter paths. We quantify the advantage of cell corners over cell centers theoretically with tight worst-case bounds on the ratios of path lengths, and empirically on a large set of benchmark test cases. We also quantify the advantage of 8-neighbor grids over 4-neighbor grids.

This paper describes a new candidate method of computing autonomous "self confidence." We describe how to analyze a plan for possible but unexpected break down cases and how to adapt the plan to circumvent those conditions. We view the result plan as more stable than the original one. The ability of achieving such plan stability is the core of how we propose to compute a system’s self confidence in its decisions and plans. This paper summarizes this approach and presents a preliminary evaluation that shows our approach is promising.

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

Just as there exists varied uses for computational models of narrative, there exists a wide variety of languages aimed at representing stories. A number of them have historic roots in automated generation, for which these languages have to be limited in order to make the generation process computationally feasible. Other are focused on story understanding, with close ties to natural language making many reasoning processes computationally intractable. In this paper, we discuss the trade-off between expressivity and computational complexity of the reasoning process and argue that Impulse, a temporal, modal logic provides more expressivity than languages historically associated with story generation, while still affording reasoning capabilities. We show that these properties enable certain aspects of narrative discourse generation by using two examples from different genres, and claim that this generalizes to a broader class of problems.

We consider stochastic planning problems that involve multiple objectives such as minimizing task completion time and energy consumption. These problems can be modeled as multi-objective Markov decision processes (MOMDPs), an extension of the widely-used MDP model to handle problems involving multiple value functions. We focus on a subclass of MOMDPs in which the objectives have a {\em relaxed lexicographic structure}, allowing an agent to seek improvement in a lower-priority objective when the impact on a higher-priority objective is within some small given tolerance. We examine the relationship between this class of problems and {\em constrained MDPs}, showing that the latter offer an alternative solution method with strong guarantees. We show empirically that a recently introduced algorithm for MOMDPs may not offer the same strong guarantees, but it does perform well in practice.

Advanced robotic technology is opening up the possibility of integrating robots into the human workspace to improve productivity and decrease the strain of repetitive, arduous physical tasks currently performed by human workers. However, coordinating these teams is a challenging problem. We must understand how decision-making authority over scheduling decisions should be shared between team members and how the preferences of the team members should be included. We report the results of a human-subject experiment investigating how a robotic teammate should best incorporate the preferences of human teammates into the team's schedule. We find that humans would rather work with a robotic teammate that accounts for their preferences, but this desire might be mitigated if their preferences come at the expense of team efficiency.

In order for robots to interact with humans in the world around them, it is important that they are not just aware of the presence of people, but also able to understand what those people are doing. In particular, interaction involves multiple agents which requires some form of coordination, and this cannot be achieved by acting blindly. The field of plan recognition (PR) studies methods for identifying an observed agent’s task or goal given her action sequence. This is often regarded as the inverse of planning which, given a set of goal conditions, aims to derive a sequence of actions that will achieve the goals when performed from a given initial state. Ram´ırez and Geffner (2009; 2010) proposed a simple transformation of PR problems into classical planning problems for which off-the-shelf software is available for quick and efficient implementations. However, there is a reliance on the observed agent’s optimality which makes this PR technique most useful as a post-processing step when some of the final actions are observed. In human-robot interaction (HRI), it is usually too late to interact once the humans are finished performing their tasks. In this paper, we describe ongoing work two extensions to make classical planning-based PR more applicable to the field of HRI. First, we introduce a modification to their algorithm that reduces the optimality bias’s effect so that long-term goals may be recognized at earlier observations. This is then followed by methods for extracting information from these predictions so that the observing agent may run a second pass of the planner to determine its own actions to perform for a fully interactive system.

Is it possible to conduct player modeling without any players? In this paper we use Monte-Carlo Tree Search-controlled procedural personas to simulate a range of decision making styles in the puzzle game MiniDungeons 2. The purpose is to provide a method for synthetic play testing of game levels with synthetic players based on designer intuition and experience. Five personas are constructed, representing five different decision making styles archetypal for the game. The personas vary solely in the weights of decision-making utilities that describe their valuation of a set affordances in MiniDungeons 2. By configuring these weights using designer expert knowledge, and passing the configurations directly to the MCTS algorithm, we make the personas exhibit a number of distinct decision making and play styles.