Purpose -The purpose of this paper is to describe a shape-shifting robot with diverse configurations, named "AMOEBA-I", which has been developed for search and rescue operations. The accessibility of this robot to unstructured environment is efficiently enhanced by changing its configuration. So the shape and reconfiguration of the robot should be considered in AMOEBA-I path planning to improve work ability of the robot in complex environment. The unique accessibility of AMOEBA-I is thus fully displayed. Design/methodology/approach -An auto-adapted path-planning method is presented for AMOEBA-I by introducing the reconfigurable ability of the robot into the modified potential field method. The modified potential field method solves the local minimum problem and goal-unreachable with nearby obstacles (GUWNO) effectively. A method of the shape-shifting robot's passing through the narrow space is studied by combining the corner detection with the modified potential field method. Findings -The ability of the robot to automatically change configuration to pass through a narrow space is proven through the experiment. Simulation results show that the robot can change its own configurations to perform auto-adapted path planning corresponding to the environmental variation. Therefore, the proposed method can improve the probability of completing the path planning. As a result, this method will shorten the path length and complete the rescue operation more effectively. Originality/value -The paper presents an effective auto-adapted path-planning method that integrates the reconfigurable ability of the robot into the modified potential field method in order to realize the auto-adapted path planning.