| One of the main goals of robotics research is to give physical platforms intelligence, allowing for the platforms to act autonomously with minimal direction from humans. Motion planning is the process by which a mobile robot plans a trajectory that moves the robot from one state to another. While there are many motion planning algorithms, this research focuses on Sampling Based Model Predictive Optimization (SBMPO), a motion planning algorithm that allows for the generation of trajectories that are not only dynamically feasible, but also efficient in terms of a user defined cost function (specifically in this research, distance traveled or energy consumed). To accomplish this, SBMPO uses the kinematic, dynamic, and power models of the robot. The kinematic, dynamic, and power models of a skid-steered robot are dependent on the type and inclination of the terrain over which the robot is traversing. Previous research has successfully used SBMPO to plan trajectories on different inclinations and terrain types, but with the terrain type and inclination being held constant over the trajectory. This research extends the prior work to plan trajectories where the terrain type changes over the trajectory and where the robot has the option to go over or around hills, situations extremely common in real world environments encountered in military and search and rescue operations. Furthermore, this research documents the design and implementation of a 3D visualization environment which allows for the visualization of the trajectory generated by the planner without having a robot follow the trajectory in a physical environment. |
