Neural Configuration Distance Function for Continuum Robot Control

2024-09-20

arXiv.org

AI Summary

This work addresses the challenge of low-fidelity shape modeling and high computational overhead in dynamic, cluttered environments, which impede real-time collision detection and safe motion planning for continuum robots. To this end, we propose the Neural Configuration-aware Euclidean Distance Function (N-CEDF)—the first approach to represent continuum robots using implicit distance fields. N-CEDF learns independent neural signed distance functions (SDFs) per segment and fuses them geometrically along the kinematic chain, enabling high-fidelity, low-latency real-time shape reconstruction. It supports millisecond-scale, point-cloud-based collision checking and is end-to-end integrated with an MPPI controller for perception-driven safe trajectory generation. In multi-segment simulations, N-CEDF improves trajectory safety by 32% and reduces planning time by 47% over baseline methods, while demonstrating robustness against both static and dynamic obstacles.

Technology Category

• Intelligent Robots: Motion and Path Planning
• Planning, Routing, and Scheduling: Mixed Discrete/Continuous Planning
• Natural Language Processing: Safety and Robustness

Application Category

• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Personalized, context-aware and across-device search

Abstract

This paper presents a novel method for modeling the shape of a continuum robot as a Neural Configuration Euclidean Distance Function (N-CEDF). By learning separate distance fields for each link and combining them through the kinematics chain, the learned N-CEDF provides an accurate and computationally efficient representation of the robot's shape. The key advantage of a distance function representation of a continuum robot is that it enables efficient collision checking for motion planning in dynamic and cluttered environments, even with point-cloud observations. We integrate the N-CEDF into a Model Predictive Path Integral (MPPI) controller to generate safe trajectories for multi-segment continuum robots. The proposed approach is validated for continuum robots with various links in several simulated environments with static and dynamic obstacles.

Problem

• Modeling continuum robot shape efficiently
• Enabling collision checking in dynamic environments
• Generating safe trajectories for multi-segment robots

Innovation

• Neural Configuration Euclidean Distance Function
• Model Predictive Path Integral controller
• Efficient collision checking for motion planning