Machine Learning for Agile and Dynamic Bipedal Locomotion with Potential Extensions to ADAS and other Automotive Safety Systems
Project Abstract/Statement of Work:
This work seeks to generate model-based feedback controllers for bipedal robots, in such a way that the closed-loop system has a very large stability basin, exhibits highly agile, dynamic behavior, and can deal with significant perturbations coming from the environment. The methods will be widely applicable beyond bipedal robots, including exoskeletons, and prostheses, and eventually, drones, ADAS and other highly automated vehicles. The skills (and eventually safety features) to be designed into the robot will be essential for robots that work around people, such as in homes or factories. In the case of bipeds: “model based” means that the controller will be designed on the basis of the full floating-base dynamic model of the robot, and not a simplified model, such as the LIP (Linear Inverted Pendulum). By “agile and dynamic” are meant that the robot moves at the speed of a normal human or faster, while stepping over or around obstacles such as a kid’s toys. By “significant perturbation” is meant a human tripping, and while falling, throwing his/her full weight into the back of the robot.
Among Hereid, Research Fellow; Divyansh Pal, Robotics MS Student; Dennis Da, ME PhD Student; and Mikhail Jones, Agility Robotics; unbox and test Cassie, EECS Prof. Jessy Grizzle’s new robot on North Campus of the University of Michigan in Ann Arbor, MI on August 22, 2017.
The robot is able to walk without a gantry and has an ankle motor that its predecessor, MARLO, lacked. The ankle motor allows for the bipedal robot to adjust more accurately to the shape and form of human movement.
Photo: Joseph Xu/Senior Multimedia Content Producer, University of Michigan – College of Engineering