Feedforward control for curve steering for an autonomous road vehicle

By the example of turning right, it is shown how autonomous mobile robots obtain additional basic skills by implementing special behavioral patterns. Autonomy is based on processing black and white monocular video image sequences in real time, supported by the 4-D approach for both vehicle state and road recognition. This approach makes use of recursive filter algorithms for estimating state variables. A feedback controller turns them into actuator signals, leading to reflex-like behavior. This mode of behavior is extended here by a rule-based one for feedforward controllers, which convert a desired pattern of behavior to actuator signals; depending on the special application to actuator signals, these may be superimposed onto those of the feedback controller, or the latter one may be suspended for some time and then newly initialized