Page 308 - Autonomous Mobile Robots

296 Autonomous Mobile Robots

and follow the leader vehicle. An autonomous tracking controller is required
for each of the follower vehicles. Based on the relative distance, orientation,
velocity, and even acceleration of the leader vehicle, the controller gener-
ates corresponding control input for the follower vehicle. Many controllers
for vehicle tracking have been proposed. In a planar configuration of vehicle
tracking, the relative position between two vehicles is basically composed of
two parts: longitudinal relative distance and lateral deviation.
Longitudinal control systems [1–5] concentrate on the longitudinal relative
distance, also called intervehicular spacing, with the assumption that the vehicle
following runs on a practically straight path or a fixed path without concerns
with steering. Thus, the tracking error is the difference between the relative
position and a predetermined spacing l. To further improve the tracking per-
formance and stability, the relative velocity of the two vehicles is also taken into
account. Using this additional tracking error, different control laws have been
proposed, for example, a simple proportional integral differential (PID) control-
ler [5] or with an additional quadratic term (PIQ controller) as in Reference 3,
and an acceleration controller with a variable feedback gain as in Reference 2;
or using adaptive control as in References 3 and 4. Lateral control, on the other
hand, is used in two applications. The first one is lane following where all
vehicles follow the center of the road or a sequence of landmarks [6–8]. The
second application, which is of our interest, concerns the path traveled by the
preceding vehicle or the leader vehicle. The only information that can be dir-
ectly measured is the relative position and orientation between two consecutive
vehicles. PID controllers are used for lateral control in References 9 and 10. A
steering controller with nonlinear feedback is presented in Reference 11. This
controller is based on a sliding mode observer and a linearized model of the
truck to issue the steering command. To achieve better performance in the lane
following method, part of the recently traveled path of the preceding vehicle
is estimated and steering control can be obtained based on linearized [12] or
nonlinear [13] dynamic/kinematic vehicle models.
Most of the above-mentioned controllers guarantee good tracking perform-
ances only when the leader vehicle moves forward in front of the follower
vehicle. Backward tracking is still a challenge due to difficulties in backward
driving as pointed out in Reference 14. Reference 14 presents a controller
that imitates the human driving of a boat with the rudder. Some preliminary
results on backward tracking for trailer systems have also been presented in
References 15–17.
Lately, a tracking control method based on output feedback theory has
been introduced in Reference 18, referred as the full-state tracking control for
wheeled mobile robots. This nonlinear tracking method ensures exponential
stability and convergence, and integrates both longitudinal control and lateral
control into one controller. In this chapter, we present a unified control design
for tracking maneuvers of two car-like mobile robots. The vehicle tracking

© 2006 by Taylor & Francis Group, LLC

FRANKL: “dk6033_c008” — 2006/3/31 — 16:43 — page 296 — #2