Perception-Driven Obstacle-Aided Locomotion for snake robots, linking virtual to real prototypes

Abstract

Snake robots have the potential to be used for many tasks where maneuvering through narrow and uneven terrain is necessary. Examples of such tasks include search-and-rescue missions, firefighting and pipe inspection. Real snakes exploit the unevenness in the terrain by strategically pushing on objects to gain forward propulsion. Research on adopting this strategy for snake robots is fairly new, and there are still many problems to be solved. In this thesis a control method for perception-driven obstacle-aided locomotion for snake robots is presented. The control method takes the multidimensional problem of controlling each joint of the robots and reduces it to a lower dimensional problem. Two main contributions were made as part of this work. The first contribution was the improvement of a previous implementation of the control method for a simulated snake robot, mainly by introducing a shape controller to aid the forward motion of the robot. The second contribution was the implementation of a communication interface between the control framework and the interface of a real snake robot. The control method was implemented in the ROS framework and tested in the robot simulator Gazebo to evaluate the performance of the design. The real snake robot was controlled through an interface in LabVIEW. The communication interface was implemented so that ROS and LabVIEW could exchange control data for both the virtual and the real snake robot. Simulations and experiments were performed to validate the implementations.