A Friction-Driven Strategy for Agile Steering Wheel Manipulation by Humanoid Robots

Area of Science:

• Robotics
• Humanoid Robot Control
• Vehicle Manipulation

Background:

• Humanoid robot maneuverability in vehicles is crucial for tasks like obstacle avoidance.
• Agile steering wheel manipulation is essential for increasing driving speed in confined spaces.
• Existing human driving strategies include 'Hand-to-Hand,' 'Hand-over-Hand,' and 'One-Hand.'

Purpose of the Study:

• To quantitatively analyze human driving strategies based on motion data.
• To propose a novel, efficient manipulation strategy for humanoid robot steering.
• To develop a control framework for precise steering and force management.

Main Methods:

• Quantitative analysis of human driving motion data (joint combination range, shoulder motion, manipulation velocity).
• Development of a friction-driven, one-hand manipulation strategy for humanoid robots.
• Construction of an operating force model for steering wheel interaction.
• Implementation of a quadratic programming (QP)-based control framework.

Main Results:

• Analysis revealed key differences in human driving strategies.
• The proposed one-hand strategy effectively balances small motion range and rapid manipulation.
• The operating force model accurately predicts steering wheel resistance.
• The QP-based controller successfully tracked end-effector position and target wrench output.
• Maximum rotation velocity achieved was 3.14 rad/s in an obstacle avoidance scenario.

Conclusions:

• The developed friction-driven, one-hand strategy significantly enhances humanoid robot steering capabilities.
• The operating force model and QP control framework enable safe and precise steering.
• This research advances humanoid robot autonomy in complex vehicle driving tasks.