A Reactive Walking Pattern Generator Based on Nonlinear Model Predictive Control

Abstract

International audienceThe contribution of this work is to show that real-time nonlinear model predictive control (NMPC) can be implemented on position controlled humanoid robots. Following the idea of " walking without thinking " , we propose a walking pattern generator that takes into account simultaneously the position and orientation of the feet. A requirement for an application in real-world scenarios is the avoidance of obstacles. Therefore the paper shows an extension of the pattern generator that directly considers the avoidance of convex obstacles. The algorithm uses the whole-body dynamics to correct the center of mass trajectory of the underlying simplified model. The pattern generator runs in real-time on the embedded hardware of the humanoid robot HRP-2 and experiments demonstrate the increase in performance with the correction

Capture-point based balance and reactive omnidirectional walking controller

“…instead of their absolute values f w i:i+3 as in [18] [20] or [21]. Similar reasoning applies also for ξ N .…”

“…Following the MPC approaches [18], [20], [21] and [14], this paper's objective can be mathematically synthesized with a cost function of the form…”

“…It allows us, unlike in [18] [21], to keep the MPC's footstep feasibility constraints linear ( (26) and (27)) except the constraints on the CoP (29). However, based on (30) and the consideration that there is no obstacle in the biped's workspace, the angular variable (θ) of the foot is free to reach, independently from the x and y variables, its prescribed value dictated by the desired rotational velocity (provided that the latter is within the robot capabilities).…”

TALOS: A new humanoid research platform targeted for industrial applications

“…The walking is therefore very reactive and can be used with a joystick. We plan to improve this algorithm by using a method coupling dynamical filter with automatic choice of foot-steps [28]. This will be done after implementing another stabilizer on the robot.…”

Model-predictive control in multi-contact based on stability polyhedrons

“…Finally, some MPC aim at generating spatial trajectories by solving non-linear problems, and then using Time Optimal Path Parametrization [13], [14], [15] to determine at which speed to execute the trajectory. Performing TOPP is fast, but solving the non-linear spatial problem can be costly unless specific assumptions or simplifications are made.…”