A multimodal approach for localization of Ackerman steering micro ground vehicles in bad GPS reception environments

A closed-loop control system requires a state estimator for feedback. In the domain of mobile robots, this is called localization. The problem of localization for micro ground vehicles in outdoor environments with no modifications to the environment is a challenging one as the tolerances for micro ground vehicles are lower. The error in the data from the GPS is higher for ground vehicles than it is for aerial vehicles in urban environments. Different approaches such as the use of optical flow and non-holonomic constraints have been explored in the past on an individual basis but have their disadvantages. In this work, these methods, along with a mathematical model of the physical system were studied individually and then combined together to eliminate their individual pitfalls. This significantly increased the reliability of the system across different scenarios.