Task planning in robotics: an empirical comparison of PDDL- and ASP-based systems

Abstract

Robots need task planning algorithms to sequence actions toward accomplishing goals that are impossible through individual actions. Off-the-shelf task planners can be used by intelligent robotics practitioners to solve a variety of planning problems. However, many different planners exist, each with different strengths and weaknesses, and there are no general rules for which planner would be best to apply to a given problem. In this article, we empirically compare the performance of state-of-the-art planners that use either the Planning Domain Description Language (PDDL), or Answer Set Programming (ASP) as the underlying action language. PDDL is designed for task planning, and PDDL-based planners are widely used for a variety of planning problems. ASP is designed for knowledge-intensive reasoning, but can also be used for solving task planning problems. Given domain encodings that are as similar as possible, we find that PDDL-based planners perform better on problems with longer solutions, and ASP-based planners are better on tasks with a large number of objects or in which complex reasoning is required to reason about action preconditions and effects. The resulting analysis can inform selection among general purpose planning systems for particular robot task planning domains.

MPTP: Motion-Planning-aware Task Planning for Navigation in Belief Space

“…However the graph of |𝑅| with planning time (Fig. 14) follows a similar trend to that reported in [33].…”

“…Their task planner is based on Answer Set Programming (ASP) [32]. Jiang et al [33] focus exclusively on task planning in robotics, assuming that a feasible motion plan exists for the synthesized task plan. They provide a comparison between ASP-based and PDDL-based task planners using different benchmark domains and conclude that PDDL-based planners perform better on tasks with long solutions, and ASP-based planners tend to perform better on shorter tasks.…”

“…While the planning time does scale exponentially with |𝑅|, the plan for |𝑅| = 9 is computed in less than 3 minutes. The work in [33] evaluates the task planning performance on a similar domain randomly selecting the number of rooms to visit in each trial. Since MPTP performs task-motion planning, the overall MPTP planning times with increasing |𝑅| is greater than those reported in [33].…”

“…The work in [33] evaluates the task planning performance on a similar domain randomly selecting the number of rooms to visit in each trial. Since MPTP performs task-motion planning, the overall MPTP planning times with increasing |𝑅| is greater than those reported in [33]. However the graph of |𝑅| with planning time (Fig.…”

The PETLON Algorithm to Plan Efficiently for Task-Level-Optimal Navigation

“…PDDL was developed for the International Planning Competition (IPC) and has been maintained by the IPC community. A recent empirical comparison has shown that PDDL-based planners perform better when tasks require long solutions, and ASP-based planners perform better when tasks require complex reasoning (Jiang et al, 2019b). The goal of implementing both ASP-based and PDDL-based planners is to provide evidence for our hypothesis that PETLON is not sensitive to task planner selection.…”