DDPG-based controlling algorithm for upper limb prosthetic shoulder joint

ABSTRACT

The development of intelligent prostheses has effectively improved the life of amputees. However, the current prosthetics mainly focus on restoring the basic mobility of amputees, without considering the use habits of the wearer and the diversity of arm movements, which makes the unknown interference and complex control requirements in daily life an obstacle to the use of prosthetics. To solve this problem, this paper proposes a combination method of adaptive control algorithms of bionic arm shoulder joint based on DDPG to realise intelligent control of the shoulder joint of upper limb prosthesis. Based on using adaptive control to reduce the interference of external variables, the accuracy of the joint module system is improved through reinforcement learning. The results show that the controller has a good effect on improving the dynamic performance of the mechanical system and can be widely used in bionic mechanical control.

KEYWORDS:

• Reinforcement learning
• bionic arm
• adaptive control
• joint module

Disclosure statement

No potential conflict of interest was reported by the author(s).