Design and development of an adaptive-torque-based proportional-integral-derivative controller for a two-legged robot

The torque-based proportional-integral-derivative (PID) controller proposed in this paper is operated on joint motors by minimizing the error between the target and actual angular displacements of that joint to achieve a superior control. On the basis of the aforementioned phenomenon, the execution effectiveness of the two-legged robot's gait can be controlled. The PID controller parameter tuning is a laborious and time-consuming process, and the controllers' performance depends on the gain values set for the controller. To overcome the said drawbacks, this study develops an adaptive-torque-based PID controller for a two-legged robot that can provide adaptive gains on the basis of the magnitude of input signal received at the input nodes of the neural network (NN). Furthermore, the structure of the feedforward NN has been optimized with the help of the modified chaotic invasive weed optimization (MCIWO) algorithm. Moreover, the concept of zero moment point has been employed to check the balance of the two-legged robot while walking on various terrains (that is, flat, stair, and sloped surfaces). The effectiveness of the developed controller has been verified in both simulations and practical.

Automated summary

This paper proposes a torque-based PID controller for a two-legged robot, aiming to optimize the gait execution. The controller adjusts gains based on the magnitude of input signals received by the neural network, and the structure of the feedforward NN is optimized using the MCIWO algorithm. The effectiveness of the controller in maintaining balance on various terrains is verified through simulations and practical experiments.