The Liner Shipping Fleet Repositioning Problem (LSFRP) poses a large financial burden on liner shipping firms. During repositioning, vessels are moved between services in a liner shipping network. The LSFRP is characterized by chains of interacting activities, many of which have costs that are a function of their duration; for example, sailing slowly between two ports is cheaper than sailing quickly. Despite its great industrial importance, the LSFRP has received little attention in the literature. We show how the LSFRP can be solved sub-optimally using the planner POPF and optimally with a mixed-integer program (MIP) and a novel method called Temporal Optimization Planning (TOP). We evaluate the performance of each of these techniques on a dataset of real-world instances from our industrial collaborator, and show that automated planning scales to the size of problems faced by industry.

We present the domain-independent HRFF algorithm, which solves goal-oriented HMDPs by incrementally aggregating plans generated by the METRIC-FF planner into a policy defined over discrete and continuous state variables. HRFF takes into account non-monotonic state variables, and complex combinations of many discrete and continuous probability distributions. We introduce new data structures and algorithmic paradigms to deal with continuous state spaces: hybrid hierarchical hash tables, domain determinization based on dynamic domain sampling or on static computation of probability distributions' modes, optimization settings under METRIC-FF based on plan probability and length. We deeply analyze the behavior of HRFF on a probabilistically-interesting structured navigation problem with continuous dead-ends and non-monotonic continuous state variables. We compare with HAO* on the Rover domain and show that HRFF outperforms HAO* by many order of magnitudes in terms of computation time and memory usage. We also experiment challenging and combinatorial HMDP versions of benchmarks from numeric classical planning.

We consider the problem of computing shortest paths subject to an additional resource constraint such as a hard limit on the (positive) height difference of the path. This is typically of interest in the context of bicycle route planning, or when energy consumption is to be limited. So far, the exact computation of such constrained shortest paths was not feasible on large networks; we show that state-of-the-art speed-up techniques for the shortest path problem, like contraction hierarchies, can be instrumented to solve this problem efficiently in practice despite the NP-hardness in general.

Human users who execute an automatically generated plan want to understand the rationale behind it. Knowledge-rich plans are particularly suitable for this purpose, because they provide the means to give reason for causal, temporal, and hierarchical relationships between actions. Based on this information, focused arguments can be generated that constitute explanations on an appropriate level of abstraction. In this paper, we present a formal approach to plan explanation. Information about plans is represented as first-order logic formulae and explanations are constructed as proofs in the resulting axiomatic system. With that, plan explanations are provably correct w.r.t. the planning system that produced the plan. A prototype plan explanation system implements our approach and first experiments give evidence that finding plan explanations is feasible in real-time.

Engineering companies working with machine tools will often be required to calibrate those machines to international standards. The calibration process requires various errors in the machine to be measured by a skilled expert. In addition to conducting the tests, the engineer must also plan the order in which the tests should take place, and also which instruments should be used to perform each test. It is critical to find as short a calibration plan as possible so that the machine is not out of service for too long. In this work, automated planning is applied to the problem of generating machine tool calibration plans. Given a description of a machine, and its various axes, we produce a calibration plan that minimises the time taken to measure all of the errors of a machine. We also consider the case when there is not enough time to test all errors of the machine, and the calibration plan must maximise the importance of the set of errors tested in the limited time available.

This paper provides an analysis of the efficacy of a known iterative improvement meta-heuristic approach from the AI area in solving the Blocking Job Shop Scheduling Problem (BJSSP) class of problems. The BJSSP is known to have significant fallouts on practical domains, and differs from the classical Job Shop Scheduling Problem (JSSP) in that it assumes that there are no intermediate buffers for storing a job as it moves from one machine to another; according to the BJSSP definition, each job has to wait on a machine until it can be processed on the next machine. In our analysis, two specific variants of the iterative improvement meta-heuristic are evaluated: (1) an adaptation of an existing scheduling algorithm based on the Iterative Flattening Search and (2) an off-the-shelf optimization tool, the IBM ILOG CP Optimizer, which implements Self-Adapting Large Neighborhood Search. Both are applied to a reference benchmark problem set and comparative performance results are presented. The results confirm the effectiveness of the iterative improvement approach in solving the BJSSP; both variants perform well individually and together succeed in improving the entire set of benchmark instances.

The paper illustrates a novel approach to conformant planning using classical planners. The approach relies on two core ideas developed to deal with incomplete information in the initial situation: the use of a classical planner to solve non-classical planning problems, and the reduction of the size of the initial belief state. Differently from previous uses of classical planners to solve non-classical planning problems, the approach proposed in this paper creates a valid plan from a possible plan---by inserting actions into the possible plan and maintaining only one level of non-deterministic choice (i.e., the initial plan being modified). The algorithm can be instantiated with different classical planners---the paper presents the GC[LAMA] implementation, whose classical planner is LAMA. We investigate properties of the approach, including conditions for completeness. GC[LAMA] is empirically evaluated against state-of-the-art conformant planners, using benchmarks from the literature. The experimental results show that GC[LAMA] is superior to other planners, in both performance and scalability. GC[LAMA] is the only planner that can solve the largest instances from several domains. The paper investigates the reasons behind the good performance and the challenges encountered in GC[LAMA].

The need to economize limited resources, such as fuel or money, is aubiquitous feature of planning problems. If the resources cannot bereplenished, the planner must make do with the initial supply. It isthen of paramount importance how constrained the problem is,i.e., whether and to which extent the initial resource supply exceedsthe minimum need. While there is a large body of literature on numericplanning and planning with resources, such resource constrainednesshas only been scantily investigated. We herein start to address thisin more detail. We generalize the previous notion of resourceconstrainedness, characterized through a numeric problem feature C≥1 , to the case of multiple resources. We implement an extendedbenchmark suite controlling C . We conduct a large-scale study of thecurrent state of the art as a function of C , highlighting whichtechniques contribute to success. We introduce two new techniques ontop of a recent Monte Carlo Random Walk method, resulting in a plannerthat, in these benchmarks, outperforms previous planners whenresources are scarce ( C close to 1 ). We investigate the parametersinfluencing the performance of that planner, and we show that one ofthe two new techniques works well also on the regular IPC benchmarks.

We address the problem of computing a policy for fully observable non-deterministic (FOND) planning problems. By focusing on the relevant aspects of the state of the world, we introduce a series of improvements to the previous state of the art and extend the applicability of our planner, PRP, to work in an online setting. The use of state relevance allows our policy to be exponentially more succinct in representing a solution to a FOND problem for some domains. Through the introduction of new techniques for avoiding deadends and determining sufficient validity conditions, PRP has the potential to compute a policy up to several orders of magnitude faster than previous approaches. We also find dramatic improvements over the state of the art in online replanning when we treat suitable probabilistic domains as FOND domains.

In contrast to previous competitions, where the problems were goal-based, the 2011 International Probabilistic Planning Competition (IPPC-2011) emphasized finite-horizon reward maximization problems with large branching factors. These MDPs modeled more realistic planning scenarios and presented challenges to the previous state-of-the-art planners (e.g., those from IPPC-2008), which were primarily based on domain determinization — a technique more suited to goal-oriented MDPs with small branching factors. Moreover, large branching factors render the existing implementations of RTDP- and LAO-style algorithms inefficient as well. In this paper we present GLUTTON, our planner at IPPC-2011 that performed well on these challenging MDPs. The main algorithm used by GLUTTON is LR2TDP, an LRTDP-based optimal algorithm for finite-horizon problems centered around the novel idea of reverse iterative deepening. We detail LR2TDP itself as well as a series of optimizations included in GLUTTON that help LR2TDP achieve competitive performance on difficult problems with large branching factors -- subsampling the transition function, separating out natural dynamics, caching transition function samples, and others. Experiments show that GLUTTON and PROST, the IPPC-2011 winner, have complementary strengths, with GLUTTON demonstrating superior performance on problems with few high-reward terminal states.