Locomotion of a Two-Dimensional Walking-Climbing Robot Using A Closed-Loop Mechanism: From Gait Generation to Navigation

Abstract

We investigate the locomotion of a surface-walking/climbing robot, the Planar Walker, based on a novel planar eight-bar closed-loop mechanism. The robot can produce inchworm-like movements in two orthogonal directions and can rotate about itself. The locomotion mechanism to achieve such motions comprises four two-way linear cylinders forming a deformable quadrilateral and four two-way gripper modules. Decoupled transverse gaits and turning gaits of the robot with finite lengths and finite rotation angles are obtained through the actuation of linear cylinders and grippers. Actuation sequences of the cylinders and grippers for different types of gaits are modeled using finite state machines. The kinematics of the gaits are described using planar rigid motion group SE(2). When a series of gaits are executed, the robot follows a non-smooth segmented trajectory. Based on this feature, three point-to-point navigation methods are developed for various scenarios: the Simple Line of Sight (SLS) algorithm, the Simulated Annealing based Accurate Planning (SAAP) algorithm, and the Localized Hybrid Accurate Planning (LHAP) algorithm. Computer simulation shows that the SAAP algorithm produces accurate gait sequences and the LHAP algorithm saves computation time and resources for long-range targets. However, experimental results show that the inherent positional errors in individual gait movements can be substantial when accumulated and can render a particular algorithm less effective.