Bioinspired preactivation reflex increases robustness of walking on rough terrain

Area of Science:

• Robotics
• Biomechanics
• Control Systems

Background:

• Bipedal robots struggle with rough terrain, unlike humans who naturally adapt to perturbations.
• Neuromusculoskeletal (NMS) models are key for transferring human control strategies to robots.
• Current reflex-based NMS models lack robustness against unexpected environmental changes.

Purpose of the Study:

• To enhance the robustness of a widely used NMS walking model by incorporating a human-like muscle preactivation strategy.
• To investigate the impact of preactivation on adapting landings and improving stability during perturbations.
• To explore the potential of this bio-inspired approach for robotic locomotion on challenging surfaces.

Main Methods:

• Integrated a muscle preactivation feedback loop into an existing reflex-based NMS walking model.
• Simulated unexpected step-down perturbations to assess model performance.
• Analyzed changes in knee kinematics and overall system robustness.

Main Results:

• The enhanced NMS model demonstrated improved adaptation to landings.
• Robustness to unexpected step-down perturbations increased significantly, from 3 cm to 10 cm.
• The stabilizing effect was attributed to altered knee kinematics resulting from preactivation.

Conclusions:

• Muscle preactivation acts as an accommodation strategy, enhancing stability without explicit perturbation detection.
• This bio-inspired reflex adaptation can significantly improve bipedal robot performance on rough terrain.
• Preactivation strategies can be ported to robotic systems to increase resilience to unexpected environmental challenges.