Optimization method of exoskeleton main assist parameters based on deep reinforcement learning

Abstract

The invention discloses an optimization method of exoskeleton main assist parameters based on deep reinforcement learning, which adopts an exoskeleton assist curve equation of composite sine curve todetermine the exoskeleton main assist parameters, and utilizes a deep deterministic strategy gradient method in the deep reinforcement learning to solve the continuity control problem of flexible exoskeleton, a strategy network and an evaluation network are established, hip joint flexion angle information of an exoskeleton wearer is collected and processed in real time to generate a data set for parameter training, training optimization of exoskeleton main assist parameters is conducted, and self-adaptive optimization of the exoskeleton main assist parameters is achieved.

Description

[0001] (1) Technical field: [0002] The present invention relates to the technical field of robots, in particular to a method for optimizing main assist parameters of an exoskeleton based on deep reinforcement learning. [0003] (two) background technology: [0004] For traditional lower limb rehabilitation training, it is guided by professional doctors and completed with the assistance of nurses or family members. This method is time-consuming, ineffective, and labor-intensive. In order to reduce the burden of manpower and achieve efficient rehabilitation services, flexible exoskeletons for gait rehabilitation have been widely used. [0005] The gait rehabilitation flexible exoskeleton combines intelligent robot technology and rehabilitation medical theory, and can replace professional doctors to help patients complete lower limb rehabilitation training. Its appearance provides a new option for the rehabilitation treatment of patients with lower limb dysfunction, and makes u...

AI Technical Summary

Problems solved by technology

Although the PID algorithm is still widely used in today's control due to its simple principle and easy adjustment of parameters, it may cause unexpected results in some cases.

For example, when the difference between the expected value and the actual value is too large, the motor will generate too high a speed to achieve the expected value, which usually leads to overshoot and oscillation, which is quite dangerous for gait rehabilitation flexible exoskeletons

Moreover, this method cannot achieve power-assisted parameter optimization, which is not only inefficient but also has large errors in parameters.

[0009] In view of the deficiencies of existing technologies, there is an urgent need for an exoskeleton main assistance parameter optimization method based on deep reinforcement learning to deal with the continuous assistance of flexible exoskeleton for gait rehabilitation

Images