A driver model with supervision aspects

In an Intelligent Transport System, for a vehicle which follows a path without the presence of a human driver, the control law development requires the modeling of the human driver behavior. A driver behavior are classified in three different levels: Control, Supervision, and Definition. (1) Control level is the lowest level with the responsibility of reaching the desired speed and maintaining the vehicle on the path. This level models driver compensatory reactions. (2) Some of the driver intelligent activities which play a role in the modeling of the driver behavior are realized at the Supervision level. This level is placed above the control level and it defines and modifies the control level desired values. (3) The itinerary, road geometric limits and speed limits are defined at the Definition level.; In this thesis the focus is on the control law developments for the control level, which requires an analysis on the kinematics and dynamics of ground vehicles. The research has been carried out in three steps which resulted in three publications given in appendices A, B and C.; The first step consists in developing a control law based on a dynamic model of car-like mobile robots. The controllability of omnidirectional mobile robots has been vastly studied before. Taking advantage of these studies, the kinematic differences between these two mobile robot families have been pointed out. Input/Output Feedback linearization (I/OFL) method has been used to develop the first control law commanding a car-like robot speed and its instantaneous curvature.; In the second step, the objective is to complete the vehicle lateral control. But, the I/OFL cannot be realized for the Cartesian coordinates of any point attached to the vehicle main body, however it is possible for a point attached to the front steerable tire. Similar to the first step, the nonlinear dynamic model has been used for the controller design.; Conventionally, the control of robots is based on inputs of desired trajectories, which are functions of time. The approach is not a convenient choice for the structure of our control level. Hence, in the third step, the lateral control law has been developed using a geometric lateral-offset as the system input instead of the conventional desired trajectories. This input removes the time dependence of lateral inputs and the vehicle speed control becomes independent from the vehicle lateral control. As a result, the supervision level can modify the control level inputs (desired values) which is an essential element for the integration of the driver intelligent activities at the supervision level.