Stochastic path generation for anticipative road following in naturally complex scene

To expand the scope of humans´ perception for anticipative road following, a stochastic perception-control coupling is introduced on satellite-roadway-vehicle network. Based on the prediction of roadway pattern in the satellite image, maneuverable area is coded in terms of a fractal attractor confined by boundary probability in the image of observed scene. Random shift of the boundary probability is detected via the perception process and interpolated by using stochastic growth dynamics. Guided by the interpolation mechanism, the control process generates the trajectory to be followed by vehicle mechanism subjected to human´s unexpectable decision making. The validity of stochastic trajectory generation is verified via simulation studies.