A task and motion planning framework using iteratively deepened AND/OR graph networks

2025-03-10

Robotics and Autonomous Systems

AI Summary

This work addresses two key challenges in task-and-motion planning (TAMP): (1) the unknown number of subtasks—e.g., dynamic determination of object re-arrangements in cluttered scenes for target grasping—and (2) the tight coupling between task assignment and motion coordination in multi-robot systems. We propose a differentiable AND/OR graph neural network framework grounded in iterative deepening search. To our knowledge, this is the first approach to integrate iterative deepening into differentiable symbolic-neural hybrid planning, enabling end-to-end joint optimization of task decomposition and motion planning—thereby overcoming the abstraction-execution decoupling inherent in conventional hierarchical architectures. Evaluated in simulation and on real robotic platforms, our method achieves a 27% improvement in planning success rate and a 41% reduction in computation time, while significantly enhancing generalization and robustness for long-horizon, dependency-rich tasks.

Technology Category

• Planning, Routing, and Scheduling: Learning for Planning and Scheduling
• Intelligent Robots: Motion and Path Planning
• Humans and AI: Human-Aware Planning and Behavior Prediction

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications

Problem

• Develops a dynamic AND/OR graph network for task and motion planning.
• Addresses unknown sub-task counts in cluttered object retrieval scenarios.
• Enables multi-robot coordination and task allocation in complex environments.

Innovation

• Uses AND/OR graph networks for task planning
• Dynamically grows AND/OR graphs at runtime
• Supports multi-robot task and motion planning