Sensor data fusion for body state estimation in a bipedal robot and its feedback control application for stable...

Area of Science:

• Robotics
• Control Systems
• Sensor Fusion

Background:

• Bipedal robots require accurate spatial motion estimation for stable locomotion.
• Integrating data from multiple sensors (joint encoders, IMU, inclinometer) is crucial for robust state estimation.
• Existing methods may not sufficiently address real-time feedback control for dynamic gaits.

Purpose of the Study:

• To develop and evaluate a sensor data fusion algorithm using an extended Kalman filter for bipedal robot spatial motion estimation.
• To utilize the estimated body state for real-time feedback control, including walking gaits.
• To implement and test a posture corrector to minimize unwanted torques during robot motion.

Main Methods:

• Implemented an extended Kalman filter for sensor data fusion from joint encoders, a 6-axis inertial measurement unit (IMU), and a 2-axis inclinometer.
• Developed a feedback control strategy incorporating the body state estimator and a damping controller for real-time position regulation.
• Integrated a posture corrector to mitigate motion-induced torques.

Main Results:

• The sensor fusion algorithm successfully estimated the bipedal robot's body state, including the center of mass (CoM) in standing posture.
• The integrated feedback control system demonstrated effective real-time regulation of the robot's body position during walking gaits.
• Experimental evaluation on a child-size bipedal robot confirmed the performance of the body state estimator and control structure.

Conclusions:

• The proposed extended Kalman filter-based sensor fusion algorithm provides accurate spatial motion estimation for bipedal robots.
• The developed feedback control strategy enhances the stability and performance of bipedal robots during walking.
• The system shows promise for practical applications requiring precise control of bipedal robots.