Hybrid Dynamic Simulation For Performance Optimization Of Legged Mechanisms

Abstract

Agile locomotion is needed for mobile robots to efficiently navigate challenging terrain. The ability of an agile legged mobile robot to abruptly change trajectory allows it toquickly react to obstacles and successfully operate in environments usually more suited forlegged animals. The research presented herein aims to increase the agility and performanceof legged robots.Using hybrid dynamic simulation, novel methods are developed to model the inter-action of a legged robot with the ground. In considering robot agility, an accurate model ofthe events that occur while in contact is needed. Methods are developed to model groundinteraction where oblique angled impacts occur, and to address the well known issues withenergy consistency when using rigid body models for dynamic systems.The contact model in this work is investigated with three multibody benchmark cases.A cable driven single leg jumping robot is modeled to research agility. An optimizationof the initial posture of the robot, and its effect on jumping performance and agility ispresented. Configuration optimization during the stance phase of a non-periodic jumpingmotion is performed, utilizing the directional dynamic capability equations. This optimization maximizes the time the mechanism is in contact with the ground, minimizes actuationeffort, and reduces the likelihood of slipping and stumbling.