Case Study: University of Michigan achieves accelerated controller development and real‑time bipedal robot control with MathWorks

University of Michigan Develops Controls for Bipedal Robots with Model-Based Design

The University of Michigan, led by Professor Jessy Grizzle and researchers including Brian Buss and Brent Griffin, faced a complex controls challenge developing MARLO, a freely moving bipedal robot with 13 degrees of freedom but only six actuators (an underactuated system). To move beyond hard‑to‑maintain hand‑coded C implementations and to prototype, verify, and deploy real‑time controllers that interface with EtherCAT hardware, the team adopted MathWorks’ Model‑Based Design tools, notably MATLAB, Simulink, and Simulink Real‑Time.

Using MathWorks software (MATLAB, Simulink, Symbolic Math Toolbox, Optimization Toolbox, Stateflow, and Simulink/MATLAB Coder), the team modeled MARLO’s dynamics, designed and optimized nonlinear controllers, generated production code, and deployed it to a Speedgoat real‑time target over EtherCAT. The MathWorks‑based workflow sped controller development by removing hand‑coding bottlenecks, enabled rapid simulation‑to‑hardware iteration, produced a controller that has allowed MARLO to walk outdoors, and has been adopted by other research groups at the University of Michigan and institutions such as MIT and Oregon State University.

University of Michigan

Jessy Grizzle

Professor