In (Otte and Correll 2013) we present C-FOREST, a parallelization framework for single-query sampling-based shortest-path planning algorithms. C-FOREST has been observed to have super linear speedup on many problems, e.g., paths of quality Ltarget are found 350X faster by 64 CPUs working in parallel than by 1 CPU. In (Otte and Correll 2013) C-FOREST is tested in conjunction with the RRT* algorithm. In the current work we perform additional experiments that show C-FOREST provides similar advantages when used conjunction with the SPRT algorithm. This reinforces our original claim that C-FOREST is generally applicable to a wide range of sampling based motion planning algorithms.

In order to overcome a large variety of run-time constraints, robots are being designed to be more resourceful by incorporating more sensory and motor options for any given task. The added flexibility provides a basis for dexterous problem solving, but challenges planners by increasing the complexity of search. Moreover, the cost of functionally equivalent options can vary dramatically. In the worst case, naive approaches to planning avoid expensive actions until inexpensive options are explored exhaustively leading to poor overall search performance. We present a dexterous robot that introduces multiple types of locomotor actions with significant differences in cost and situational value and apply standard search techniques to demonstrate the additional challenges that arise in the context of dexterous mobility. Results highlight incentives, opportunities, and impact for overcoming these challenges. Additionally, we present a prototype for a path planner that uses environmental features to define an efficient set of subgoals for dexterous motion planning.

Due to the high complexity of probabilistic planning algorithms, roboticists often opt for deterministic replanning paradigms, which can quickly adapt the current plan to the environment's changes. However, probabilistic planning suffers in practice from the common misconception that it is needed to generate complete or closed policies, which would not require to be adapted on-line. In this work, we propose an intermediate approach, which generates incomplete partial policies taking into account mid-term probabilistic uncertainties, continually improving them on a gliding horizon or regenerating them when they fail. Our algorithm is a configurable anytime meta-planner that drives any sub-(PO)MDP standard planner, dealing with all pending and time-bounded planning requests sent by the execution framework from many reachable possible future execution states, in anticipation of the probabilistic evolution of the system. We assess our approach on generic robotic problems and on combinatorial UAVs (PO)MDP missions, which we tested during real flights: emergency landing with discrete and continuous state variables, and target detection and recognition in unknown environments.

We present an algorithm for summarizing the metric resource requirements of an abstract task based on the resource usages of its potential refinements. We use this summary information within the ASPEN planner/scheduler to coordinate a team of rovers that conflict over shared resources. We find analytically and experimentally that an iterative repair planner can experience an exponential speedup when reasoning with summary information about resource usages and state constraints, but there are some cases where the extra overhead involved can degrade performance.

As modeling details can have a large impact on planner perormance, domain transformation has been a traditional subject of interest in the planning community not only betweenlanguages, but also within languages. Herein, we automatean intra-language transformation method that has as yet beenapplied only manually, and that has never been formally described: action schema splitting, which transforms an actionschema with a big interface (many parameters) into severalschemas with smaller interfaces, exponentially reducing thenumber of ground actions. We spell out this method, characterizing exactly the choice of splits preserving equivalence tothe original schema. Making that choice involves a trade-off between interface size and plan length, which we explore bydesigning automatic domain optimization methods. Our experiments show that these methods can substantially improveperformance on domains with large interfaces.

There is a strong demand for robots to work in environments, such as aircraft manufacturing, where they share tasks with humans and must quickly adapt to each other's needs. To do so, a robot must both infer the intent of humans, and must adapt accordingly. The literature to date has made great progress on these two tasks - recognition and adaptation - but largely as separate research activities. In this paper, we present a unified approach to these two problems, in which recognition and adaptation occur concurrently and holistically.  Key to our approach is a task representation that uses choice to represent alternative plans for both the human and robot, allowing a single set of algorithms to simultaneously achieve recognition and adaptation. To achieve such fluidity, a labeled propagation mechanism is used where decisions made by the human and robot during execution are propagated to relevant future open choices, as determined by causal link analysis, narrowing the possible options that the human would reasonably take (hence achieving intent recognition) as well as the possible actions the robot could consistently take (adaptation). This paper introduces Pike, an executive for human-robot teamwork that quickly adapts and infers intent based on the preconditions of actions in the plan, temporal constraints, unanticipated disturbances, and choices made previously (by either robot or human).  We evaluate Pike's performance and demonstrate it on a household task in a human-robot team testbed.

We examine new ways to perform plan recognition (PR) using natural language processing (NLP) techniques. PR often focuses on the structural relationships between consecutive observations and ordered activities that comprise plans. However, NLP commonly treats text as a bag-of-words, omitting such structural relationships and using topic models to break down the distribution of concepts discussed in documents. In this paper, we examine an analogous treatment of plans as distributions of activities. We explore the application of Latent Dirichlet Allocation topic models to human skeletal data of plan execution traces obtained from a RGB-D sensor. This investigation focuses on representing the data as text and interpreting learned activities as a form of activity recognition (AR). Additionally, we explain how the system may perform PR. The initial empirical results suggest that such NLP methods can be useful in complex PR and AR tasks.

The operational traffic control problem comes up in a number of different contexts. It involves the coordinated movement of a set of vehicles and has by and large the flavor of a scheduling problem. In trying to apply scheduling techniques to the problem, one notes that this is a job-shop scheduling problem with blocking, a type of scheduling problem that is quite unusual. In particular, we will highlight a condition necessary to guarantee that job-shop schedules can be executed in the presences of the blocking constraint. Based on the insight that the traffic problem is a scheduling problem, we can derive the computational complexity of the operational traffic control problem and can design some algorithms to deal with this problem. In particular, we will specify a very simple method that works well in fast-time simulation contexts.

In classical planning, the polynomial-time computability of propositional delete-free planning (planning with only positive effects and preconditions) led to the highly successful Relaxed Graphplan heuristic. We present a hierarchy of new computational complexity results for different classes of propositional delete-free HTN planning, with two main results: We prove that finding a plan for the delete-relaxation of a propositional HTN problem is NP-complete: hence unless P=NP, there is no directly analogous GraphPlan heuristic for HTN planning. However, a further relaxation of HTN planning (delete-free HTN planning with task insertion) is polynomial-time computable. Thus, there may be a possibility of using this or other relaxations to develop search heuristics for HTN planning.

Planning under partial observability in nondeterministic domains is a very significant and challenging problem, which requires dealing with uncertainty together with and-or search. In this paper, we propose a new algorithm for tackling this problem, able to generate conditional plans that are guaranteed to achieve the goal despite of the uncertainty in the initial condition and the uncertain effects of actions. The proposed algorithm combines heuristic search in the and-or space of beliefs with symbolic BDD-based techniques, and is fully amenable to the use of selection functions. The experimental evaluation shows that heuristic-symbolic search may behave much better than state-of-the-art search algorithms, based on a depth-first search (DFS) style, on several domains.