Technical University of Munich (TUM) and Siemens Technology
Tuesday, April 05, 2022, 11:00 - 12:00
This talk presents the results of a master thesis project carried out at the Technical University of Munich and in cooperation with Siemens. Zhang et al.  introduced a method to derive exact differentiable collision constraints for optimal control by utilizing dual formulations. Thereby, the non-differentiability of the primal version of the constraints is avoided. We use the method to derive multiple variations of such differentiable constraints and analyse their numerical performance using the CasADi framework . To this end, multiple robotic systems are considered for numerical case studies: starting with the model used in the original paper, a bicycle-based car model, we also discuss the examples of a drone and a robotic manipulator. Furthermore, different initialization strategies are evaluated in order to deduce the most efficient overall approach for the considered tasks of robotic trajectory planning.
 J. A. E. Andersson, J. Gillis, G. Horn, J. B. Rawlings, and M. Diehl. Casadi: a software framework for nonlinear optimization and optimal control. Mathematical Programming Computation, 11:1–36, 2019. https://doi.org/10.1007/s12532-018-0139-4.
 X. Zhang, A. Liniger, and F. Borrelli. Optimization-based collision avoidance, 2018. http://arxiv.org/abs/1711.03449 arXiv:1711.03449.
Meeting-ID: 627 9173 7415
The slides of the talk can be found here: syscop_pres.pdf