Abstract: Bending Pneumatic Artifi cial Muscles (PAMs) are particularly attractive and extensively applied to the soft grasper, snakelike robot, etc. To extend the application of PAMs, we fabricate a Multi-directional Bending Pneumatic Artifi cial Muscle (MBPAM) that can bend in eight directions by changing the pressurized chambers. The maximum bending angle and output force are 151° and 0.643 N under the pressure of 100 kPa, respectively. Additionally, the Finite Element Model (FEM) is established to further investigate the performance. The experimental and numerical results demonstrate the nonlinear relationship between the pressure and the bending angle and output force. Moreover, the eff ects of parameters on the performance are studied with the validated FEM. The results reveal that the amplitude of waves and the thickness of the base layer can be optimized. Thus, multi-objective optimization is performed to improve the bending performance of the MBPAM. The optimization results indicate that the output force can be increased by 7.8% with the identical bending angle of the initial design, while the bending angle can be improved by 8.6% with the same output force. Finally, the grasp tests demonstrate the grip capability of the soft four-fi nger gripper and display the application prospect of the MBPAM in soft robots. 

Key words: Pneumatic artifi cial muscle, Multi-directional bending, Soft robots, Parameter optimization