In this paper, we propose a longitudinal and lateral control of a convoy of autonomous vehicles to follow a desired trajectory and maintain an inter-distance between them to avoid collisions and realize its missions in the required condition. The control approach belongs to the decentralized global family and is based on the states estimated By different sliding mode approaches, which allow speed and position states estimations (longitudinal, lateral and yaw angle) by using position sensors, in order to calculate the dynamics of each vehicle. The convoy treated in this paper is composed of four vehicles in which a firstorder sliding mode observer (FOSM) is used for the 1 st vehicle, and a first-order sliding mode observer with a linear correction term (FOSML) for the 2 nd vehicle. The 3 rd and 4 th vehicle use successively a secondorder sliding mode observer without and with a linear correction term (SOSM, SOSML). The law of control is calculated by using a linearized control by inverse dynamics for the longitudinal movement and a firstorder sliding mode control for the lateral movement. The performance and efficiency of the proposed approaches were tested and validated in two different trajectory scenarios and with low and high speeds in the SCANeR T M-Studio platform.