Human-like hopping in machines : Feedback- versus feed-forward-controlled motions

Area of Science:

• Robotics
• Biomechanics
• Control Systems

Background:

• Template models like the spring-loaded inverted pendulum (SLIP) are crucial for analyzing legged locomotion and informing robotic designs.
• Controlling vertical hopping in legged robots presents challenges in achieving human-like performance and adaptability.

Purpose of the Study:

• To compare feed-forward and bio-inspired virtual model control strategies for vertical hopping in a 2-segmented legged robot using the SLIP model.
• To evaluate the effectiveness of a bio-inspired feedback control approach that emulates a virtual spring between the hip and foot.

Main Methods:

• Implementation of a spring-loaded inverted pendulum (SLIP) model for controlling vertical hopping.
• Comparison of traditional feed-forward control with a novel bio-inspired virtual model control incorporating feedback.
• Analysis of robot hopping dynamics and performance metrics under different control strategies.

Main Results:

• The bio-inspired virtual model control demonstrated improved hopping control compared to feed-forward methods.
• Robot hopping exhibited similarities to human hopping patterns.
• The feedback control approach enhanced perturbation recovery and locomotion adaptation, proving easier to tune.

Conclusions:

• A simple, template-based controller, specifically the SLIP model with bio-inspired feedback, can achieve human-like hopping in robots.
• This control strategy offers significant improvements in performance, robustness, and versatility for legged locomotion.
• The findings suggest a viable pathway for developing more adaptable and efficient legged robots inspired by biological systems.