Unstructured, dynamic, untidy: These are usually environments that robots don't necessarily like. Because for obstacles that suddenly appear, a strategy is needed on how to avoid a collision. For example, if a robot has the task of filling a cup with water and someone places an object or waves a stick in the air between the robot's gripper and the cup, it must be able to recognize these obstacles, immediately avoid them and find a new path using the human guidance provided in the beginning. "No matter what obstacle is placed in the way: The robot arm calculates a new path," says VDE Bayern Award winner Benjamin Bogenberger, explaining his solution. "Like a magnet with two positive poles", the robot navigates past the obstacles to ultimately be able to perform its task – for example, filling a cup with water or stacking blocks by using a single human demonstration from the human.
Bachelor thesis at the Munich Institute of Robotics and Machine Intelligence (MIRMI): Robot autonomously decides on a new route using human demonstration
The special feature of the solution, which was developed in collaboration with scientists Riddhiman Laha and Dr. Luis Figueredo from the Munich Institute of Robotics and Machine Intelligence (MIRMI): Bogenberger's new algorithm can automatically maintain distance from obstacles, immediately recognises when new obstacles are added or the overall picture changes, and autonomously decides which route to take to complete its task. The robot can therefore "generalize" the current situation, and is flexible and adaptable. "The solution is transferable to any other situation," explains Bogenberger, who discovered his affinity for robotics even before his studies. At the age of 14, he built a robot that collected balls on a tennis court. Currently, the 22-year-old Master's student is studying at the Technical University in Delft, the Netherlands, and is supporting the BMW car company as a working student for autonomous driving.Information
User Guided Motion Generation with Reactive Whole-Body Collision Avoidance for High-DoF Systems, 2022
In the picture: The winner of the VDE Award Benjamin Bogenberger (middle) and his advisors Riddhiman Laha (right) and Dr. Luis Figueredo