Capture Point-Based Controller Using Real-Time Zero Moment Point Manipulation for Stable Bipedal Walking in Human...

Area of Science:

• Robotics
• Control Systems
• Human-Robot Interaction

Background:

• Human-robot collaboration in shared environments requires robots with adaptable locomotion.
• Bipedal robots need robust stability control to navigate dynamic and unpredictable settings.
• Existing methods often lack flexibility for real-time adjustments in walking patterns.

Purpose of the Study:

• To propose a stability control method for dynamically modifiable bipedal walking.
• To enable bipedal robots to adjust walking patterns in real-time during locomotion.
• To facilitate stable human-robot collaboration in shared environments.

Main Methods:

• Utilizing a capture point (CP) tracking controller for dynamic walking.
• Generating a reference CP trajectory via real-time zero moment point (ZMP) manipulation.
• Implementing a ZMP-based CP tracking controller with force-sensing resistors for ZMP measurement.
• Incorporating a method to handle infeasible footstep commands to maintain stability.

Main Results:

• Demonstrated stable bipedal walking with dynamically modifiable patterns, including during the single support phase.
• Successfully tracked reference CP trajectories generated without future footstep information.
• Validated the method's effectiveness through simulations and experimental results.
• Ensured continuity in walking patterns during the double support phase.

Conclusions:

• The proposed CP tracking control method enables stable and adaptable bipedal robot locomotion.
• Real-time modification of walking patterns, even with sudden command changes, is achievable.
• The method enhances the potential for seamless human-robot collaboration in dynamic environments.