This page is currently under construction. Please check back soon. (Last update September 2015)
High Speed Running Control of Quadrupedal Robots
The main objective of the study is to present a controller design scheme which provides a robust running gait of a quadruped robot with the ability to change the running speed over a wide range while handling variations in ground height and stiffness. Drawing on trends in change of stance and swing duration, and vertical impulse of quadrupedal animals as speed increases, a novel algorithm is proposed to prescribe vertical and horizontal impulses to achieve the desired stance and swing duration for different speeds as well as to regulate body pitch dynamics. Application of the control design scheme not only enables MIT Cheetah 2 to run with a wide range of speeds from 0 to 4.5 m/sec (10.1 miles per hour) without change of any control parameters or a further optimization process but also provides 40 cm high dynamic jumping followed by safe landing while running with a speed of 2.5 m/sec (5.5 miles/hour). The proposed impulse planning approach is extended to the design of high jumping, long distance leaping and rapid turning motions where the jumping height, leaping distance, and turning radius can be precisely controlled as planned.
Control of a Compliant Bipedal Robotic Walker and Runner
This work was directed toward addressing restrictions on robustness of bipedal robot walking by pursuing a novel control design for MABEL, a robot testbed with compliant series springs for energy efficiency and robustness. In order to enable MABEL to walk on uneven ground without a priori information, humans’ two adaptation strategies to unexpected ground height variations are encoded: the adjustment of leg stiffness in response to unexpected changes in ground heights and reflex strategies to recover from tripping over obstacles. To encode these strategies, the framework of active force control which varies effective leg stiffness and damping to adapt to ground height variations is integrated with the method of virtual constraints. As a result of the study, a feedback controller was designed, providing experimental results of MABEL accommodating various types of platforms without a priori knowledge of the obstacles, including ascending a 12.5 cm high platform, stepping off from a 18.5 cm high platform equaling 18.5% of leg length, and walking over a platform with multiple ascending and descending steps.