1: \begin{abstract}

2: 

3: Optimization fabrics are a geometric approach to real-time local

4: motion generation, where motions are designed by the composition

5: of several differential equations that exhibit a desired motion behavior.

6: We generalize this framework to dynamic scenarios and non-holonomic robots

7: and prove that fundamental properties can be conserved. We show that convergence to

8: desired trajectories and avoidance of moving obstacles can be guaranteed using

9: simple construction rules of the components. Additionally, we present the

10: first quantitative comparisons between optimization fabrics and

11: model predictive control and show that optimization fabrics can generate

12: similar trajectories with better scalability, and thus, much

13: higher replanning frequency (up to 500 Hz with a 7 degrees of

14: freedom robotic arm). Finally, we present empirical results on

15: several robots, including a non-holonomic mobile manipulator with 10

16: degrees of freedom and avoidance of a moving human,

17: supporting the theoretical findings.

18: 

19: \changed{The open-source implementation can be found at

20:   \url{https://github.com/tud-amr/fabrics}

21: }

22: \end{abstract}

23: