Different coating technological requirements of automobile electro-coating control system have different desired trajectories. For a dual parallel robot used for automobile electro-coating conveying (DPRAEC), an intelligent tracking smooth sliding mode synchronization control method (ITSSMSC) is proposed to achieve better tracking control performance with high synchronization and flexibility for the system under different desired trajectories. According to the dual parallel and symmetrical structure of the robot, the sliding surface is designed based on a composite error. A deep neural network with multi-hidden layer is trained by deep learning algorithms. Then, any given trajectories caused by different coating technological requirements could be tracked without tuning additional parameters and the synchronous performance could be guaranteed. The convergence of the network training process and the stability of the ITSSMSC are proved theoretically. Compared with the synchronization proportion-derivative control, the sliding mode control based on constant velocity reaching law without synchronization control and the smooth sliding mode synchronization control without training, simulation and experimental results demonstrate the effectiveness of the proposed approach. Compared with synchronization proportion-derivative control, the maximum tracking errors of constant velocity reaching law without synchronization control among each active joint are, respectively, reduced 92.2%, 94.7%, 98.8%. The maximum tracking errors of smooth sliding mode synchronization control and ITSSMSC in [Formula: see text] direction are, respectively, 9.06×10−5m and 1.01×10−9m. The maximum tracking errors of smooth sliding mode synchronization control and ITSSMSC in [Formula: see text] angle are, respectively, 4.50×10−3rad and 1.45×10−4rad. Therefore, the proposed control method could improve the tracking control performance and synchronization performance simultaneously under different desired trajectories.