Technology for Creating Digital Twins to Optimize the Design Parameters of Robots and Their Control Systems

摘要

A universal software technology for rapid creation and testing of digital twins of robotic systems with adaptive control capability has been developed. It allows for flexible and relatively rapid creation of digital twins of mobile and anthropomorphic robots. A digital twin consisting of simulation models of the kinematic system, control system, and operating environment is capable of almost completely simulating the robot’s beha­ vior in various modes in real or pseudo-real time. The technology operation process is described, starting from creating a solid model and ending with selecting the optimal robot motion control method. Particular attention is paid to adaptive control through reinforcement learning, which allows the system to adapt to changing conditions. Depending on the goals of the proposed object and the availability of initial data, the proposed technology makes it possible design and implement a robot control system or test new control methods. A practical example of applying the technology for creating digital twins to control an anthropomorphic robot aimed at simulating human walking has been implemented. The results confirm a reduction in development time and an increase in the reliability of solutions. The technology is integrated into a single software package, which simplifies the creation of new twins and testing of algorithms.