Abstract: Pneumatic soft robots are suitable for performing tasks in unstructured and complex environments due to good flexibility and safety.As it is difficult to mount sensors on the soft robot to directly measure shape and posture,making it difficult to achieve accurate position control, its motion accuracy is generally much lower than that of conventional rigid robots. A vision-based method of shape reconstruction,posture measurement and inverse kinematic control of soft robot was proposed to address these issues.The experimental results show that the root-mean-square error of the end-effector position of the soft robot using the method is less than 6 mm,the position error of the key point measured by this method is less than 2.80 mm,the position deviation of the overall shape is 1.67 mm,the average errors of the bending angle and the deflection angle are 3.6° and 3.4° respectively,the maximum position error and angle error of the control reaching the desire are 3.47mm and 0.24° respectively.The experimental results show that the proposed posture measurement and control method is useful for improving the motion accuracy and working ability of the soft robot.

Key words: pneumatic soft robot, shape reconstruction, posture measurement, inverse kinematic control