This year, Nanjing's intelligent bus system has achieved remarkable results, and citizens have clearly felt that routes are more convenient and traffic is faster. Recently, the Nanjing Bus Group announced the launch of the second phase of the bus network upgrade, working with Ant Digital to create a "Bus Network Optimization Intelligent Agent." This is the first time that "AI Intelligent Agent" technology has been deeply applied to the public transportation scheduling system in China. Based on multidimensional data analysis such as stations, passenger flow, and routes, it realizes a comprehensive upgrade from "experience-based decision-making" to "intelligent planning." Problems such as "adding a few buses during the evening rush hour" and "how to adjust the route" will now have more efficient solutions. Among them, the Nanjing Bus Line 210, which was recommended by AI, serves nearly 50% elderly passengers.
Figure: Nanjing Bus Intelligent Agent Xiao Lan Jing
According to the information, since 2022, Nanjing Bus has started the intelligent upgrading of the operation scheduling system. In 2025, the "Network Optimization and Passenger Flow Prediction AI Large Model" (hereinafter referred to as "Xiao Lan Jing") co-built by Nanjing Bus and Ant Digital has achieved phased results. Xiao Lan Jing can intelligently analyze the travel time of each route from the starting point to the destination, accurately depict the city-wide, all-time, and all-mode travel map; it can accurately analyze the passenger flow rules within 200m of subway stations, bus stops, and community hubs, capture the "capillary blockage points" in travel, and generate optimal solutions for connecting routes intelligently.
A relevant person in charge of Nanjing Bus said: "In the past, many route designs were judged based on experience and needs, such as adding hospital express lines or shuttle buses for the last metro trains. However, while considering humanized aspects, we also need data statistics to verify, making decisions more accurate." Currently, Nanjing Bus has used the large model to assist in opening more than 30 micro-loop routes and added 84 pick-up points. Among them, the Bus Line 210, which was recommended by AI, greatly reduces transfer time and walking distance. According to statistics, the line had a maximum daily passenger volume of 2,168 times, with an average of 380 people served per vehicle per day, and nearly 50% of the elderly cards. "In addition to serving daily commuters, this line covers multiple tourist spots such as Xiaguan Wharf and Caihong Bridge, improving the tourists' transportation experience. It maintains the same passenger volume on weekends as on weekdays, which can be considered an 'unexpected gain' brought by AI," said a relevant person in charge of Nanjing Bus.
It is reported that Phase II of Xiao Lan Jing will introduce Ant Digital's AI Intelligent Agent technology, aiming to make bus scheduling more intuitive and efficient, and improve the smoothness and predictability of passenger travel experiences. The "Bus Intelligent Agent" of Xiao Lan Jing is based on advanced technical capabilities such as multimodal recognition, semantic understanding, logical reasoning, knowledge enhancement, and Agent tool scheduling. It can complete tasks such as detecting blind spots in bus routes and predicting passenger flow through natural language instructions, and output visually the route planning scheme, station layout plan, and economic feasibility report. In terms of underlying algorithms, Phase II will adopt a combination of a large bus model and vertical small models, developing new algorithms for specific scenarios such as micro-loop buses, bus-metro connections, and route operation optimization, to optimize analysis effects specifically.