Shape Sensing, Mapping, and Obstacle Avoidance of the Continuum Manipulator in Confined Environments

Accurate shape sensing and environmental perception are essential for the precise control and safe autonomous operation of continuum manipulators. This paper presents an integrated system combining shape sensing, environment modeling, and obstacle avoidance for confined‐space operation. For shape sensing, cylindrical capacitive sensors and an inertial measurement unit are fused, and a backpropagation neural network is used to reconstruct the manipulator configuration. The Morris method is further employed to evaluate the sensitivity of soft sensor signals and IMU orientation data to the end‐effector position. For environmental perception, a miniaturized ranging module based on VL6180X laser sensor array is developed. Through coordinate transformation, local distance measurements are mapped into the world coordinate system to construct a local environment map. Autonomous obstacle avoidance is achieved using predefined safety thresholds and a reverse motion strategy. Experiments on a single‐segment manipulator show mean position errors of 0.29 and 1.24 mm for straight‐line and circular trajectories, corresponding to relative errors of 0.31% and 1.32%. Under a 110 g tip load, the three‐segment manipulator achieves a mean end‐effector prediction error of 3.74 mm (relative error of 1.04%). The ranging module also reconstructs a 120‐mm‐diameter pipe, demonstrating the feasibility of autonomous continuum manipulator operation in confined environments.