Nonlinear model predictive control-based trust region-SQP method for gait optimization of biped robot

Abstract

The invention relates to the field of the gait optimization control for a biped robot and particularly relates to a model predictive control-based trust region-SQP method for the gait optimization of the biped robot. Based on the nonlinear model predictive control technology, the trust region-SQP method is adopted to realize the superlinear convergence rate and the global convergence fast-solving optimal control, so that the optimal movement gait of the biped robot is realized. Since the nature of an objective function is poor, a significant amount of time is spent in determining a searching step length. Therefore, the real-time performance is invalid. According to the technical scheme of the invention, based on the nonlinear model predictive control technology, a robot dynamics model is converted into a non-linear optimization model, and the trust region-SQP method having fast convergence characteristics is provided. In this way, the optimal control, which realizes the real-time performance, is solved. The method overcomes the defect that the real-time performance cannot be realized during the solving process of a traditional controller. The method can be applied to multi-degree-of-freedom biped robots and provides references for the real-time control solutions of multi-degree-of-freedom biped robots.

Description

technical field [0001] The invention relates to a gait optimization control method in the field of biped robots, in particular to a trust region-SQP method for biped robot gait optimization based on nonlinear model predictive control. [0002] technical background [0003] In recent years, more and more countries have incorporated robot research into their national development plans, such as the National Strategic Plan for Advanced Manufacturing in the United States, the SPARC Civilian Robot Research and Development Program in the European Union, Made in China 2025, New Robot Strategy in Japan, and Robotics in South Korea. "Future Strategy 2022" and others have incorporated the development of robots into the key research fields of national technological innovation and industrial development. Robots have become the focus of industrial competition around the world under the background of a new round of technological revolution and industrial transformation. At the intersection...

AI Technical Summary

Problems solved by technology

For example, the optimal control theory based on bionics not only requires a large number of experimental studies, but the control method designed from the perspective of bionics also uses the action data or biological structure of real people. However, the body mass distribution and The geometric dimensions are slightly different from those of the biped robot, which leads to a great difference in the dynamics-kinematics mathematical models of humans and robots. Applying this method directly to the actual robot system control process can lead to deviations in the system, so , destroying the dynamic and stable walking of the robot

Although the controller designed based on the idea of ​​hybrid zero dynamics realizes the dynamic and stable walking of the biped robot, this method needs to know the precise dynamics mathematical model of the biped robot in advance, otherwise, the method will fail, thus causing the robot to fail. Possibility to achieve stable periodic motion

The dynamic and stable walking of biped robots can be realized by using discrete mechanics and optimal control theory. However, this method cannot control the robot system in real time under changing working conditions.

However, as a method for real-time prediction and control, nonlinear model predictive control technology has strong robustness and adaptability, and it can solve the problem that existing control methods do not have real-time performance.

However, when the nonlinear model predictive control technology is used to discretize the nonlinear dynamic model, the dynamic model is often transformed into a non-convex optimization problem, which increases the computational complexity of the algorithm. Therefore, it brings a huge challenge to the real-time control of the system.

[0005] To sum up, the existing gait optimization control methods are difficult to ensure the real-time control of the biped robot system by the controller, so it is impossible to achieve an efficient and stable human-like walking gait

Images