Using a distributed reactive algorithm to control an arbitrary number of collaborative hyper-redundant serial manipulators in real time
This thesis presents a novel method of controlling the end effector pose of a hyper-redundant serial manipulator. Designated Briareos, the algorithm utilizes a distributed swarm-style algorithm embedded into each link in the system. Such an approach allows a highly redundant system to interact dynamically with its environment and achieve a goal without relying on limited pre-computed solutions, complex monolithic solvers, or computationally burdensome numerical search methods. This level of adaptability allows the system to operate in complex environments with obstacles. Additionally, this method is shown to extend to systems of collaborative manipulators with a shared workspace. The algorithm was implemented in simulation and rigorously tested, as well as shown to be stable and reliable by analytical methods. Two related algorithms, named Cottus and Gyes, are also introduced as solutions to the grasping problem.