In the Internet economy, the takeaway has become a popular way of consumption. However, the current takeaway route optimization model and algorithm do not consider the rider’s goal and the disturbance factors they faced, which makes the rider trapped in the system. Most studies optimize the takeaway route as a static problem, generating routes by pairwise insertion of pickup and delivery nodes. However, the takeaway route optimization is dynamic and real-time, and the cross of pickup and delivery is the staple mode of delivery, which means that riders can go to multiple nodes to pick up before delivery. Therefore, this paper studies the delivery route optimization under the cross of pickup and delivery and various interference factors. Firstly, the empty cost of riders and the cost of riders waiting are added to the target function, and an optimization model for takeaway delivery routes is established. Secondly, four interference factors are considered, including mid-way orders, traffic control, the abnormal delivery time of merchants, and the abnormal time of customer pickup. An improved adaptive large neighborhood search algorithm is designed to achieve efficient route optimization. Finally, a simulation example is generated based on the Ele.me platform to verify the effectiveness of the model and algorithm.