Contact-based Grasp Control and Inverse Kinematics for a Five-fingered Robotic Hand

2025-03-24

AI Summary

Addressing the challenge of achieving stable grasping with five-fingered robotic hands, this paper proposes a synergistic control method integrating contact force feedback and inverse kinematics. A real-time closed-loop control system is implemented in PyBullet using the DexHand v2 model: contact point optimization and force-closure verification ensure grasp stability, while an optimization-driven inverse kinematics solver enhances joint motion efficiency and pose accuracy. The approach uniquely unifies high non-thumb phalangeal motion efficiency (0.966–0.996), low positioning error (≤0.0283 m), and high-frequency real-time convergence (240 Hz); for the thumb, efficiency reaches 0.879 with an error of 0.0519 m. This work pioneers the deep coupling of high-fidelity contact-force closed-loop control with fine-grained inverse-kinematic resolution, significantly improving robustness and responsiveness during complex object manipulation.

Technology Category

• Intelligent Robots: Manipulation
• Search and Optimization: Sampling/Simulation-based Search
• Humans and AI: Human-Aware Planning and Behavior Prediction

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
• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments

Abstract

This paper presents an implementation and analysis of a five-fingered robotic grasping system that combines contact-based control with inverse kinematics solutions. Using the PyBullet simulation environment and the DexHand v2 model, we demonstrate a comprehensive approach to achieving stable grasps through contact point optimization with force closure validation. Our method achieves movement efficiency ratings between 0.966-0.996 for non-thumb fingers and 0.879 for the thumb, while maintaining positional accuracy within 0.0267-0.0283m for non-thumb digits and 0.0519m for the thumb. The system demonstrates rapid position stabilization at 240Hz simulation frequency and maintains stable contact configurations throughout the grasp execution. Experimental results validate the effectiveness of our approach, while also identifying areas for future enhancement in thumb opposition movements and horizontal plane control.

Problem

• Implementing contact-based grasp control for five-fingered robotic hands
• Optimizing inverse kinematics for stable grasping with force closure
• Improving movement efficiency and positional accuracy in robotic grasping

Innovation

• Combines contact control with inverse kinematics
• Uses PyBullet and DexHand v2 simulation
• Optimizes contact points with force closure