Efficient Trajectory Optimization for a Free-Floating Robot

Kurzfassung

Trajectory planning for complex systems like free-floating robots requires much computation. In this thesis, two applications, namely a trajectory optimization algorithm and a method for generating maps of the singular configuration space of such a system, are improved in efficiency. The trajectory optimziation benefits from exploiting the locality property of nonuniform B-Splines, which are used to parametrize parts of the trajectory. This lead to improvements of up to 5.78 in our applications. It is also shown in this thesis, that the speed improvement directly depends on the problem at hand. Singularity maps are generated via taylor model arithmetics, which relies on an efficient implementation, which was lacking in previous work. The implementation resulting from this work is 30 to 70 times faster than the previous implementation, allowing more complex setups in the future.