Learn2Decompose: Learning Problem Decomposition for Efficient Sequential Multi-object Manipulation Planning

2024-08-13

AI Summary

Traditional task-and-motion planning (TAMP) solvers suffer from exponential growth in replanning time as problem scale increases, hindering real-time performance in dynamic, multi-object sequential manipulation tasks. Method: This paper proposes a demonstration-based problem decomposition framework that innovatively integrates three components: goal decomposition learning, temporal distance prediction, and object reduction. It combines behavioral cloning with state-sequence modeling to autonomously identify subgoals, and synergistically employs heuristic object filtering and symbolic reasoning for lightweight replanning. Contribution/Results: The resulting hybrid TAMP paradigm significantly reduces replanning latency across three standard benchmarks, enabling real-time response in long-horizon, multi-object scenarios—achieving average speedups of several-fold. It is the first approach to deliver efficient, interpretable hierarchical task-motion replanning under environmental disturbances.

Technology Category

• Planning, Routing, and Scheduling: Replanning and Plan Repair
• Humans and AI: Human-Aware Planning and Behavior Prediction
• Intelligent Robots: Motion and Path Planning

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

We present a Reactive Task and Motion Planning (TAMP) approach for efficient sequential multi-object manipulation in dynamic environments. Conventional TAMP solvers experience an exponential increase in planning time as the planning horizon and number of objects grow, limiting their applicability in real-world scenarios. To address this, we propose learning problem decomposition from demonstrations to accelerate TAMP solvers. Our approach consists of three key components: goal decomposition learning, temporal distance learning, and object reduction. Goal decomposition identifies the necessary sequences of states that the system must pass through before reaching the final goal, treating them as subgoal sequences. Temporal distance learning predicts the temporal distance between two states, enabling the system to identify the closest subgoal from a disturbed state. Object reduction minimizes the set of active objects considered during replanning, further improving efficiency. We evaluate our approach on three benchmarks, demonstrating its effectiveness in improving replanning efficiency for sequential multi-object manipulation tasks in dynamic environments.

Problem

• Efficient sequential multi-object manipulation planning in dynamic environments
• Exponential planning time growth in conventional TAMP solvers
• Learning problem decomposition to accelerate TAMP solvers

Innovation

• Learning problem decomposition from demonstrations
• Predicting temporal distance between states
• Minimizing active objects during replanning