Soft Robotics
"Adaptive Control of a Soft Quadruped Robot with Pneumatic Artificial Muscles for Bio-Inspired Locomotion"
Role: Graduate Research Assistant
In my master's research, I developed a bio-inspired soft quadruped robot modeled after the musculoskeletal structure of a greyhound, using 30 McKibben-type pneumatic artificial muscles (PAMs). These actuators were antagonistically arranged to mimic natural joint function across 4-DOF limbs. I designed custom PAMs using rubber tubing and nylon braided sleeves, and fabricated bone structures based on greyhound anatomical proportions.
To control the robot, I implemented a solenoid valve system with CAN-based communication and feedforward control via dual Arduino Dues. I designed and tested six muscle activation patterns and gait cycles, adjusting the stance and swing phase durations to optimize locomotion. Motion capture tools were used to perform biomechanical analysis on joint trajectories, stride length, and step height.
As a result, I achieved stable adaptive 3D walking and improved the robot’s gait performance. The walking speed increased by 81% (from 0.11 m/s to 0.20 m/s), and walking stability improved by 25%. Muscle activation precision was enhanced by 30% through real-time valve control and optimized timing. This research provides insight into biologically grounded soft locomotion systems and sets a foundation for future tension-feedback-based control.
Research Outputs:
Invited Talks:
1. You Wear It Well Podcast: Design of a bio-inspired musculoskeletal soft quadruped robot, Nov. 2023
Research Support:
Scholarship:
This research is supported by the Japanese Government (Monbukagakusho) Scholarship that is funded by the Ministry of Education, Culture, Sports, Science and Technology of Japan.