Shape-changing robots show great potential in locomotion on complex terrains for their capability of reshaping to fit the rough surfaces and to move omnidirectionally. In previous studies, many kinds of shape-changing robots have been proposed, and manual discrete locomotion planning method was adopted in most cases. However, autonomy is highly demanded in field working, especially when the robot is confronted with varied circumstances. In this work, we focus on autonomous locomotion planning for the planar closed-loop robot. Physics-driven locomotion planning method is proposed based on modelling an artificial dynamical process, which is intended to generate feasible configurations subject to the local terrain. To verify our methodology, we developed a polygonal robot prototype and conducted both dynamic simulations and laboratory experiments. The results show that the proposed method is capable of generating feasible locomotion control scheme for various working conditions, such as curve terrain, obstacle and step terrain.

变形机器人在复杂地形上表现出巨大的运动潜力，因为它们具有重新塑形以适应粗糙表面和全方位移动的能力。在以前的研究中，已经提出了多种变形机器人，并且在大多数情况下采用手动离散运动计划方法。但是，在现场工作中，特别是在机器人遇到各种情况时，对自治性的要求很高。在这项工作中，我们专注于平面闭环机器人的自主运动计划。提出了一种基于人工动力过程建模的物理驱动运动计划方法，旨在根据当地地形生成可行的配置。为了验证我们的方法，我们开发了一个多边形机器人原型，并进行了动态仿真和实验室实验。