Combining high-level causal reasoning witth low-level geometric reasoning and motion planning for robotic manipulation

Official URL: http://192.168.1.20/record=b1378260 (Table of Contents)

We present a modular planning framework for manipulation tasks that combines high-level representation and causality-based reasoning with low-level geometric reasoning and motion planning. This framework features bilateral interaction between task and motion planning, and embeds geometric reasoning in causal reasoning. The causal reasoner guides the motion planner by finding an optimal task-plan; if there is no feasible kinematic solution for that task-plan then the motion planner guides the causal reasoner by modifying the planning problem with new temporal constraints. The geometric reasoner guides the causal reasoner to find feasible kinematic solutions by means of external predicates/functions. We show the applicability of this method on two sample problems: extended towers of Hanoi and multiple robot manipulation inside a maze. We focus on two main problems in this planning framework: i) a systemic analysis of various levels of integration between high-level representation and causality-based reasoning with low-level geometric reasoning and motion planning and ii) generalization of the planning framework to continuous domains. For the former, we consider various levels of integration in the two domains mentioned above, to check which level of integration achieves better performance. For the latter, we abstract configurations at the representation level by continuous regions instead of discrete positions, and introduce an incremental sampling-based method coupled to a goal region-based probabilistic path planner for extracting specific goal configurations required for generating valid plans for execution. This way, we tightly integrate high-level reasoning and region-based motion planning and provide a general framework for addressing a wide spectrum of manipulation problems.