Tokyo police to introduce AI-based accident reconstruction system

Tokyo’s Metropolitan Police Department plans to introduce a new system that can reconstruct traffic accidents using artificial intelligence, it was learned Wednesday.

The system will combine AI with 3D spatial data from accident sites and footage from nearby security cameras, reducing the need for lengthy on-site inspections, which require traffic restrictions and often cause congestion and economic losses.

The MPD plans to test the new system from June 2027 and launch full-scale operations in January 2028. Related expenses have been included in the police department’s budget request for the next fiscal year.

Currently, fatal accidents and hit-and-run incidents are investigated through on-site inspections, in which vehicles are positioned as they were at the time of the crash and measurements are taken to produce reports and diagrams.

Such inspections sometimes require traffic restrictions lasting two to three hours, resulting in congestion. In some cases, more than 20 personnel are mobilized, including traffic controllers.

According to the MPD, the new system will utilize an existing mobile mapping system in which camera-equipped vehicles capture surroundings to generate 3D spatial data.

The system will allow investigators to extract necessary information, such as vehicle speed, distance, and positional relationships, on a screen.

It will also enable simulations from multiple perspectives, including overhead views of collisions and the viewpoints of drivers and pedestrians, making it easier to prepare investigative reports and diagrams.