Optimal configuration control of planar leg/wheel mobile robots for flexible obstacle avoidance. (August 2020)

• Record Type: Journal Article
• Authors: Takahashi, Naoki; Shibata, Naoki; Nonaka, Kenichiro
• Is Part Of: Control engineering practice. Volume 101(2020)
• Publication Date: 2020-08
• Subjects: Planar leg/wheel mobile robot -- Model predictive control -- Leg configuration -- Obstacle avoidance; Automatic control -- Periodicals; 629.89
• DOI: 10.1016/j.conengprac.2020.104503
• Languages: English
• ISSNs: 0967-0661

Abstract

Abstract: In this study, we propose a method that can be used to enhance the configuration of mobile robots equipped with articulated legs and wheels by simultaneously optimizing leg joint angles and wheel positions using model predictive control (MPC). To present a manageable expression for MPC, the kinematics of a robot that has highly articulated legs with wheels at their ends are represented as serially connected constrained links. The other end of the leg is constrained by its position relative to the robot body. To actively allocate and adapt movements to the surrounding environment, we use a potential field method applied to each wheel. Our method is based on the following strategy. First, we will design a simple collision-free reference path for a point mass, and then we will apply MPC to induce optimal path tracking control that balances the input and artificial potential costs. This will allow the robot to adapt its wheel position configurations to the surrounding environment. Based on that strategy, we conducted numerical simulations and experiments using an actual mobile robot to verify the efficacy and feasibility of our proposed method, the results of which demonstrated its effective ability to adaptively configure its articulated legs in ways that allowed the robot to avoid obstacles. Graphical abstract: Highlights: Active leg configuration during obstacle avoidance is proposed. Enhanced leg configuration is achieved using model predictive control. Kinematics of articulated leg is represented as serially connected constraints. Experiments is executed to demonstrate the efficacy of our proposed method.