Towards Real-Time Parameter Optimization for Feasible Nonlinear Control with Applications to Robot LocomotionReport as inadecuate


Towards Real-Time Parameter Optimization for Feasible Nonlinear Control with Applications to Robot Locomotion


Towards Real-Time Parameter Optimization for Feasible Nonlinear Control with Applications to Robot Locomotion - Download this document for free, or read online. Document in PDF available to download.

This paper considers the application of classicalcontrol methods, designed for unconstrained nonlinear systems,to systems with nontrivial input constraints. As shown throughoutthe literature, unconstrained classical methods can be usedto stabilize constrained systems, however, without modificationthese unconstrained methods are not guaranteed to work for ageneral control problem. In this paper, we propose conditionsfor which classical unconstrained methods can be guaranteedto exponentially stabilize constrained systems – which we term-feasibility- conditions – and we provide examples of howto construct explicitly feasible controllers. The control designmethods leverage control Lyapunov functions CLF describingthe -desired behavior- of the system; and we claim that in theevent that a system’s input constraints prevent the productionof an exponentially stabilizing input for a particular CLF, anew, locally feasible CLF must be produced. To this end, wepropose a novel hybrid feasibility controller consisting of acontinuous-time controller which implements a CLF and adiscrete parameter update law which finds feasible controllerparameters as needed. Simulation results suggest that theproposed method can be used to overcome certain catastrophicinfeasibility events encountered in robot locomotion.



Advanced Mechanical Bipedal Experimental Robotics Lab AMBER Publications - Advanced Mechanical Bipedal Experimental Robotics Lab AMBER -



Author: Powell, Matthew J. - Ames, Aaron D. - -

Source: https://smartech.gatech.edu/







Related documents