Abstract: In the automotive industry, optimizing the production process of manufacturing identical products is the key to improving production efficiency. Robot automated production lines are an ideal choice for modern manufacturing. The existing research mainly focuses on the finite subproblems of production lines. This paper considers the welding task assignment and welding sequence planning of multi-station and multi-robot production line in the actual automobile manufacturing industry, in which many constraints of workpiece and production line composition are involved. A mathematical model of Multi-station Multi-robot Welding Task Assignment and Path Planning (MSMR-WTAPP) is established, whose optimization objective is to minimize the first workpiece processing time, the subsequent workpiece processing time, and the robot motion path length synchronously. An improved SPEA2 algorithm (SPEA2+IPD) based on Individual population density (IPD) is proposed and a dual layer coding scheme is designed for the coupling problem of welding task assignment and welding sequence planning, and the decoding process of robot operation is addressed. Finally, the effectiveness and superiority of SPEA2+IPD algorithm in optimizing the cycle time, efficiency and welding path of multi station and multi robot production lines are verified via simulation experiments. Compared to the actual production cycle of the factory, the production cycle optimized by SPEA2+IPD algorithm is 20.7% shorter for the first workpiece, and the production time of each subsequent workpiece is reduced by 15.2%. This indicates that the proposed model and algorithm have practical significance for optimizing the factory production.