Collaborative trajectory generation for a group of mobile manipulators. Distributed optimization, RRT-based algorithm and online collision avoidance
Supervisor: Prof. Oliver Sawodny
Advisor: Alice Hierholz
Automation of interior construction through mobile manipulators offers a promising avenue to increase productivity and resource efficiency in the construction sector. This thesis investigates collaborative trajectory planning methods for groups of mobile manipulators. lt analyzes distributed optimization approaches based on ADMM and leader-follower strategies, as weil as centralized RRT-based planning methods. While distributed optimization approaches are constrained by limited real-time capability, the RRT algorithm in combination with Bezier splines shows potential for quasi-real-time applications. In addition, online collisionavoidance methods-particularly control barrier functions-are explored to locally secure pre-planned trajectories. The focus is on precise and safe motion planning, collision avoidance, and coordinated cooperation between robots. Cooperative transport of a wooden beam du ring studwall construction in interior construction is chosen as an example application. The proposed methods are first validated on a simplified drivingdynamics model and then partially applied to a high-dimensional robot model. Overall, this work contributes to automated, collaborative motion planning of mobile manipulators in the construction industry.
