Work on the methods of optimal control methods for multibody and robotic systems

Methods for modelling and simulation of the dynamics of complex robotic systems understood as multibody systems constitute a mature field, especially in the context of sequential low-order algorithms, but this does not mean that the field is complete and there are no areas that require improvement.

The main objective of the project conducted by the team at the Faculty of Power and Aeronautical Engineering, Warsaw University of Technology, was to develop and implement efficient methods for modelling multibody and robotic systems unified with non-direct nonlinear optimal control methods (adjoint).

“Through the use of such algorithms, we generate control signals and shape a nominal motion trajectory that prompts the robot's control system," explains Professor Pawel Malczyk, head of the project. “This task has been solved for many years in robotics. In the project, we focused on optimising the mission objectives, such as avoiding obstacles, minimising energy consumption, improving the quality of control during the robot’s movement, as well as issues related to increasing the efficiency of computation by using parallel processing. We have implemented the solutions developed in the control system of a parallel robot with two degrees of freedom, but the devised methods are universal enough to be applied to the control of other multibody systems,” he adds.

The project was successful, and all the outcomes assumed in the project were achieved. “In addition, I wanted to involve young people in this project, and this also worked because I was joined by two doctoral students in the research team," says Professor Malczyk.

-

The “Efficient indirect optimal control methods for multibody systems” project was financed as part of the research grant under the “Excellence Initiative – Research University” programme implemented at the Warsaw University of Technology.

Research team:
Professor Paweł Malczyk; Paweł Maciąg, MSc; Maciej Pikuliński, MSc.