Related Experiment Videos
Benchmark evaluation in task and motion planning using iteratively deepened AND/OR graph networks
Hossein Karami1, Antony Thomas2, Fulvio Mastrogiovanni1
1Department of Informatics, Bioengineering, Robotics, and Systems Engineering, University of Genoa, Genoa, Italy.
Frontiers in Robotics and AI
|July 30, 2026
Summary
This study introduces a robust task-motion planning (TMP) framework. The iterative deepening AND/OR graph-based planner successfully addresses diverse challenges across five standardized benchmarks in robotics.
Area of Science:
- Robotics
- Artificial Intelligence
- Motion Planning
Background:
- Robotics research involves complex challenges across various subdomains.
- Standardized benchmarks are crucial for systematic and comprehensive evaluation of robotics frameworks.
- Existing applications may not encompass the full spectrum of challenges within a domain.
Purpose of the Study:
- To evaluate the performance of a novel task-motion planning (TMP) framework.
- To demonstrate the framework's effectiveness across diverse and challenging benchmarks.
- To validate the robustness of the proposed TMP solution.
Main Methods:
- Description of an iterative deepening AND/OR graph-based task-motion planning (TMP) planner.
- Performance assessment of the TMP framework on five community-proposed benchmarks.
- Evaluation methodology focused on addressing specific challenges inherent to each benchmark.
Main Results:
- The proposed TMP planner successfully solved all five evaluated benchmarks.
- The framework demonstrated effectiveness in handling diverse aspects of task-motion planning.
- Consistent performance across benchmarks indicates a robust and generalizable solution.
Conclusions:
- The developed task-motion planning (TMP) framework is a robust and effective solution.
- The framework's success across multiple benchmarks validates its capability in addressing complex robotics challenges.
- This work contributes a reliable approach to task-motion planning in robotics research.