This work proposes a novel control system that enables highly reconfigurable robots to traverse confined truly 3D environments (such as those encountered during Urban Search and Rescue operations). This is achieved by three significant advancements, i) by breaking the complete motion planning problem down into smaller sub-problems each of which is easier for the planner to solve, ii) by utilizing a novel search algorithm, RAIN, which combines aspects of systematic sampling, random sampling, and gradient ascent style search techniques, and iii) by using a novel mapping algorithm which combines spatial occupancy mapping and neural networks.

The new control system is then used to overcome multiple basic obstacles commonly used to test current Urban Search and Rescue robots. Test results indicate that the control system is equally capable of overcoming both simple and complex obstacles without dependency of pre-selected gaits or environmental simplification.