Learning a Kinodynamic Trajectory Manifold for Impact-Aware Compliant Catching of Fast-Moving Objects

2026-05-27

AI Summary

This work addresses the challenges of short reaction time, high collision uncertainty, and stringent dynamic constraints in high-speed aerial object capture by proposing a real-time trajectory generation method that eliminates the need for online nonlinear optimization. The approach leverages reinforcement learning in simulation to collect successful grasping trajectories and constructs a low-dimensional trajectory manifold of the manipulator’s dynamics, enabling direct mapping from the object’s initial state to a reference trajectory. By integrating impact-awareness with compliant control, the method achieves precise tracking and effective impact absorption near contact. Experimental results demonstrate that the proposed framework significantly improves capture success rate, dynamic adaptability, and overall robustness under uncertain impact conditions.

Abstract

Fast catching of free-flying objects is difficult because of short reaction time, impact uncertainty, and kinodynamic constraints. We use reinforcement learning in simulation to collect successful catching trajectories and learn a low-dimensional kinodynamic trajectory manifold. At run time, the estimated object initial state is mapped directly to a reference catching trajectory without online nonlinear optimization. The trajectory is tracked with compliant control near contact for improved impact absorption and capture stability.

Problem

• kinodynamic constraints
• impact-aware catching
• fast-moving objects
• compliant control
• trajectory planning

Innovation

• kinodynamic trajectory manifold
• reinforcement learning
• compliant catching
• impact-aware control
• real-time trajectory mapping