Artificial Intelligence/Machine Learning to Improve Maneuver of Robotic/Autonomous Systems

Robotic autonomous systems (RAS) are currently being used for many different applications using a wide variety of vehicle platforms. The environments in which RAS are being used are becoming increasingly complex. Vehicle path planning and control is challenging in environments with many obstacles and uneven terrain. This proposed research will develop and compare multiple techniques to improve vehicle response through difficult terrain. The current state-of-the-art will be researched to find methods that can be used to improve vehicle maneuverability in difficult terrain. Two other methods will be compared. These methods are nonlinear model predictive control (NMPC) and a machine learning method associated with trajectory generation. This work will compare these methods in simulation and then implement and test the best method on a fully automated Ford Escape.