Tracking Control for Multiple Trailer Systems by Adaptive Algorithmic Control

In recent years, a truck-trailer system is the most useful physical distribution system. The truck-trailer systems have more convenience than coastal services or freight trains. Meanwhile, problems of the traffic jam and the air pollution in an urban area have become serious, year after year. Therefore improvement and rationalization of the transport efficiency are social needs. There are many papers suggesting a platoon system of several trucks as a part of development of ITS (Intelligent Transport System). These platoon systems consist of several unmanned trucks automatically following a truck driven by a conductor, and it is commonly believed that it brings improvements of energy efficiency along with alleviation of the traffic jam. Moreover, there is a purpose of covering insufficient workforce of truck drivers who have to do severe labors, too. In the platoon, trucks are not physically connected to each other, and thus there is much flexibility. On the other hand, even if each vehicle is physically connected by mechanical linkage, this is not important restrictions, for transport robots which are operated in the factory, because moving range and action plan are limited. Moreover, the multiple trailer system is safer than platoon system, because if each vehicle is physically connected, there is no danger of collision among trailers. In this paper, we deal with a control method for a physically connected multiple trailer robot, which is a transport system in factories. The control method of connected vehicle has been studied for a long time (Laumond, 1986). In particular, there are many papers which studied controlling its backward motion with guaranteed stability (Sampei & Kobayashi, 1994). Moreover, kinematical model of a multiple trailer system is described by a nonholonomic system, and it is a controllable nonlinear system (Hermann & Krener, 1977). In theoretical field, it has been a hot subject of research, because asymptotic stabilization is impossible using one continuous time-invariant since the nonholonomic system does not satisfy the Brockett's necessary condition for stabilizability (Brockett, 1983). Therefore, the control problem of nonholonomic system is a theoretically difficult problem, thereupon various researches such as time-variant controller (M'Closkey & Murray, 1993) or hybrid control techniques (Matsune et al., 2005) are performed. We look at this issue from more practical point of view, then investigate a real-time control algorithm, which is based on the so called algorithmic control (Kobayashi et al., 2005a), (Imae et al., 2005) with a similar formulation of the model predictive control (MPC)