The centroid of the whole vehicle moves when the automatic guided vehicle (AGV) loads and unloads goods between stations in the intelligent factory, which reduces the trajectory tracking accuracy. To this end, the dynamics and kinematics models of a four-wheel steering AGV were established, and the Lyapunov direct method was used to construct a trajectory tracking controller with global asymptotic stability in this study. Based on the adaptive learning factor and inertia weight, an adaptive particle swarm optimization algorithm was designed to optimize the control parameters of the controller, and an adaptive global asymptotic tracking control (AGATC) controller was proposed. Under simulated working condition of moving centroid, the AGATC controller was compared with adaptive model predictive control (AMPC) controller, and the trajectory tracking simulation was carried out. The results show that the position deviation of the AGATC controller was 23.97% lower than that of the AMPC controller, and the trajectory tracking control accuracy is higher under the condition of moving centroid. Moreover, a prototype of AGV was developed, and the trajectory tracking control verification experiment was carried out. The results show that the error between simulation and experiment was less than 9.03%, which proves the authenticity and effectiveness of the AGATC controller. This study provides theoretical and experimental basis for intelligent factory to realize precision and intelligent handling technology.

The centroid of the Automated Guided Vehicle (AGV) equipped with the load platform is high, and it is prone to rollover when driving sideways on a slope. To solve this problem, this study proposes an anti-rollover cooperative control strategy combining differential drive, active steering, and load platform. According to the characteristics of AGV, the dynamic coupling of each subsystem is analyzed. The expected yaw stability control torque is calculated based on optimal control. The active steering controller is designed based on the model predictive control. The expected roll control torque is solved by the proposed sliding mode variable structure control method to dynamically adjust the centroid. Evaluation of the overall control system is accomplished by simulations and experiments under different load conditions of the AGV in the lateral driving condition on a slope. Compared with the PI controller, the sideslip angle decreases by 14.62% and 18.24%, and the roll angle decreases by 12.38% and 14.93% under high and low load conditions, respectively. The error between the experimental and simulation results is within 7.8%. It shows that the proposed cooperative control strategy can improve the stability of AGV under different load conditions and reduce the rollover probability when the AGV is driving sideways on a slope. This research provides a theoretical and experimental basis for active safety control of AGVs.