A Model-Based Control Approach for Locomotion of Biped Robots

In this research we aim at proposing a general novel walking method for locomotion of torque controlled robots. The method should be able to produce a wide range of speeds without requiring off-line optimizations and re-tuning of parameters. It should be capable of tolerating internal errors, noises and control delays as well as external disturbances such as pushes or roughness in the environment. We have a quadratic whole-body optimization which generates joint toques, given desired Cartesian accelerations of center of mass and feet. Using dynamics model of the robot inside this optimizer ensures compliance and better tracking, required for fast locomotion. We have simplified the model of robot to linear inverted pendulum and proposed different planners which are other quadratic convex problems optimizing future behavior of the robot. These planners are in fact model predictive control which optimize the system either in continuous or discrete time domains. Fast libraries help us performing these calculations per time step and producing desired motion. With very few parameters to tune and no perception, our method shows notable robustness against strong external pushes, large terrain variations, internal noises, model errors and also delayed communication. Evident by various simulations in different conditions, we can suggest our general method for walking control of a wide range of humanoid robots.