Development of a self-righting behavior on the four-legged walking robot CHARLIE

 One important aspect in the field of legged robotics is robustness to unforeseen events during a mission. While walking on rugged terrain, a legged robot is prone to fall. In order to continue its mission, the robot needs to provide a self-righting behavior, which succeeds starting from as many as possible poses.
A self-righting behavior has to provide joint trajectories that bring the robot back into a standing pose, from which it can continue its operation. Challenges during generation of trajectories are the contextual dependency of the initial pose and the often large configuration space of robots with a high number of degrees of freedom.
This thesis targets to develop different self-righting behaviors and compare them in terms of robustness, energy efficiency and development effort. At first, an engineered approach was realized and then optimized using Evolutionary Strategies. Another approach will be derived by Evolutionary Strategies with minimized prior knowledge.
The simulation model of the four-legged robot Charlie is used as a development and test platform. Charlie is biologically inspired by an ape. Its features are different four legged walking gaits, an artificial spine and different proprioceptive and interoceptive sensors.