<p>This paper addresses the formation trajectory tracking control problem for networked multi-agent systems with time delays and data loss. Firstly, an improved artificial potential field based path planning algorithm is proposed to overcome unreachable target and local minimum in traditional methods, ensuring convergence of the virtual leader to the target and smoother obstacle avoidance trajectory. Secondly, a networked predictive control scheme is introduced to solve communication constraints, and the future state of every follower is predicted by utilizing outdated state information to compensate for time delays and data loss. Thirdly, the predictive-based control protocol is proposed to guarantee that all followers maintain the desired formation while tracking the obstacle avoidance trajectory planned by the virtual leader. Finally, numerical simulations are provided to demonstrate the effectiveness of the proposed approaches, by comparing with the traditional artificial potential field method and the protocol with delayed information.</p>