Load-adaptive Servo Control On Humanoid Joints

A humanoid robot belongs to robots with serial joints. When the robot is in motion, theload torque and the rotary inertia of joint motors may vary according to joint angles. Inorder to improve the control performance of a humanoid robot and the load adaptability ofjoint motors, a load-adaptive controller is developed. The main content is as follows.First, the paper analyzed the joint servo system of a humanoid robot with variableloads. The paper introduced the servo control system of BHR5, established the single jointcontrol model with double closed loops based on a direct current motor, and analyzed howthe changing load torque and rotary inertia affect joint servo systems in theory. And further,the effects on dynamic performance and static performance of servo systems were analyzedthrough simulation.Second, a load-adaptive PID control algorithm is proposed. After analyzing theadvantages and shortcomings of PID control with auto-tuning parameters based on BPneural network, it is presented that a PID controller based on neural network offline tuning.Several kinds of PID parameters variation models including impaired, invariant, andenhanced, were designed according to the different feedback error state. And the variationfactors reflecting the variability of the PID parameters were set by the tuning results of theBP neural network. The method takes full advantage of the self-learning of the neuralnetwork, avoids its weakness such as complex algorithm, long learning time and so on, andimproves the variable load adaptability of the servo system. In addition, the hardware andsoftware systems of the joint controller were designed, and the presented algorithm isrealized through software programming.At last, simulation and experiment on the load-adaptive servo system are conducted.The servo performances were made a comparison under the control of the proposed neuralnetwork PID controller and the traditional PID controller. The results of simulation andexperiment showed that the designed joint servo system had good load adaptability,improving the servo capability and decreasing the response errors.