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.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: in robot task planning, select an appropriate task planning system (based on PDDL or ASP) to deal with different types of planning problems. Specifically, the author hopes to help robot task planning practitioners understand the impact of domain - specific properties on planning time through an empirical comparison of the performance of two different planning languages (PDDL and ASP), thereby providing guidance for choosing a suitable planning language. ### Main research questions 1. **Differences in the performance of PDDL and ASP in different types of tasks**: - PDDL (Planning Domain Definition Language) is mainly used for task planning. - ASP (Answer Set Programming) was originally designed for knowledge - intensive reasoning, but can also be used for task planning. 2. **How to select an appropriate planning language according to task characteristics**: - For tasks that require long - term solutions, which planning language performs better? - For tasks with a large number of objects or requiring complex reasoning, which planning language is more effective? ### Research background - **History of task planning**: Task planning is one of the oldest research fields in artificial intelligence. Since STRIPS (Fikes and Nilsson, 1971), many languages for representing task planning domains have been developed. - **Characteristics of PDDL and ASP**: - PDDL is a widely - used task planning language that supports logical axiom reasoning. - ASP is non - monotonic and is suitable for tasks that require default reasoning and inertial fact reasoning. ### Experimental design - **Benchmark problems**: Three different benchmark problems were selected for comparison: - **Robot Navigation**: Involves complex action pre - conditions and effect reasoning. - **Blocks World**: The classic block - stacking problem. - **Hiking**: Involves the arrangement of multiple couples' hiking trips. - **Experimental setup**: - Use FastDownward and Clingo as solvers. - Compare the planning time in different - scale domains (small domains and large domains). - Change domain characteristics (such as the number of rooms, block height, etc.) to measure the performance differences of different planning paradigms. ### Conclusions - **Situations where PDDL performs better**: - In tasks that require long - term solutions, PDDL performs excellently. For example, in the Robot Navigation domain, as the number of rooms to be visited increases, the growth of PDDL's planning time is relatively gentle. - **Situations where ASP performs better**: - In tasks with a large number of objects or requiring complex reasoning, ASP performs better. For example, when the number of rooms is small, ASP's planning speed is significantly faster than that of PDDL, and it is more efficient in handling complex reasoning. ### Hypothesis verification - **Hypothesis 1**: The current state - of - the - art PDDL - based planners perform better in tasks that require long - term solutions. - Verification result: Confirmed in all three benchmark domains. - **Hypothesis 2**: ASP - based planners perform better in tasks that require complex reasoning. - Verification result: Confirmed in the control experiment, especially in the Robot Navigation and Blocks World domains. Through these empirical comparisons, this paper provides valuable references for robot task planning practitioners, helping them make more appropriate choices when facing different types of planning problems.

Task Planning in Robotics: an Empirical Comparison of PDDL-based and ASP-based Systems

iCORPP: Interleaved commonsense reasoning and probabilistic planning on robots

An Optimal Task Planning and Agent-aware Allocation Algorithm in Collaborative Tasks Combining with PDDL and POPF

Robot Planning Based on Behavior Tree and Knowledge Graph

Robot Task Planning Based on Large Language Model Representing Knowledge with Directed Graph Structures

Logic programming for deliberative robotic task planning

Robot Task Planning and Situation Handling in Open Worlds

Exhaustiveness Does Not Necessarily Mean Better: Selective Task Planning for Multi-robot Systems.

Simultaneous Task Allocation and Planning for Multi-Robots under Hierarchical Temporal Logic Specifications

Integrating Robot Task Planner with Common-sense Knowledge Base to Improve the Efficiency of Planning

Task Roadmaps: Speeding up Task Replanning

LaMMA-P: Generalizable Multi-Agent Long-Horizon Task Allocation and Planning with LM-Driven PDDL Planner

Combined Task and Motion Planning as Classical AI Planning

CLMASP: Coupling Large Language Models with Answer Set Programming for Robotic Task Planning

Optimal Robotic Assembly Sequence Planning: A Sequential Decision-Making Approach

Knowledge-Based Hierarchical POMDPs for Task Planning

Describe, Explain, Plan and Select: Interactive Planning with LLMs Enables Open-World Multi-Task Agents.

Human-Aware Robot Task Planning Based on a Hierarchical Task Model

plasp 3: Towards Effective ASP Planning

Describe, Explain, Plan and Select: Interactive Planning with Large Language Models Enables Open-World Multi-Task Agents