Eventos Anais de eventos
COBEM 2021
26th International Congress of Mechanical Engineering
Driving Motions for Wheeled-Legged Robots in Rough Terrain Using 2D Trajectory Optimization
Submission Author:
Vivian Suzano Medeiros , RJ , Brazil
Co-Authors:
Vivian Suzano Medeiros, Marco Antonio Meggiolaro
Presenter: Vivian Suzano Medeiros
doi://10.26678/ABCM.COBEM2021.COB2021-0908
Abstract
A field of research that has received a lot of funding in the last few years is the development of mobile robots, autonomous or remotely operated, that can perform tasks in rough terrain. In most of these applications, hybrid wheeled-legged robots are well-suited since they combine the high mobility of the legs with the efficiency of the wheels. For such robots, driving motions are particularly interesting, because they allow for faster and more stable locomotion, and obstacle negotiation is still possible through the presence of the legs. When planning for driving motions, common scenarios often include trajectories in terrains with little variation in the lateral direction. Such scenarios can benefit from a simplification of the planning problem and can be handled with a two-dimensional approach. In this context, we present a formulation of the 2D Trajectory Optimization (TO) problem for quadrupedal wheeled-legged robots driving in rough terrain that optimizes over the robot's state and ground reaction forces in a single planar problem. The formulation uses the map of the terrain to optimize the driving motions while taking into account the dynamics of the robot. The main advantage of this approach is the fast planning times, which are achieved by representing the robot with a simplified two-dimensional single rigid body model that still ensures physically feasible motions. The optimization is formulated as a Nonlinear Programming Problem (NLP) and solved using a fast interior-point solver. The suitability of several cost functions is evaluated for the approach in conditions of fast locomotion and high acceleration. The trajectories are validated in simulations with a quadrupedal robot equipped with non-steerable wheels in different challenging scenarios.
Keywords
Wheeled-legged robots, Trajectory optimization, Optimal Control, rough terrain, 2D model
