GM develops collective intelligence for car cameras, letting vehicles share what they see
A patent application from General Motors and the University of Michigan describes a system that would combine camera, radar, lidar and location data from multiple cars into a shared road map. The aim is to reduce blind spots and show vehicles hazards their own sensors cannot yet see.
Cars could benefit from each other’s sensors
Patent application US 20260133049 A1 is titled “Collaborative Perception System for Creating a Bird’s Eye View Cooperative Perception Map”. It was published on 14 May 2026, with the University of Michigan System and GM Global Technology Operations LLC listed as applicants.
The idea sounds simple enough. Making it work is the hard part. A car would no longer depend only on its own cameras and sensors. Instead, data collected by several vehicles would be processed into a bird’s eye view format and merged by central computers into one shared perception map.
Such a system could help in situations where a driver or assistance system cannot see a hazard directly. A large lorry, for example, might hide an approaching car or pedestrian at a junction. If another connected vehicle already sees the danger, the system could pass that information on.
The map would return to the cars
Vehicles would gather information from cameras, radar, lidar and positioning sensors. That data would travel to a cloud based system, where it would be synchronised and turned into a three dimensional picture of the traffic situation. The finished image could then be sent back to vehicle displays or driver assistance control units.
The patent description also mentions restoring missing data, cross attention and temporal attention maps. In plainer terms, the system must piece together incomplete information from different cars and turn it into a picture reliable enough to be useful.
The practical value: blind spots, pedestrians and parking
GM’s system could give a driver or automated driving system information about objects outside the vehicle’s own sensor range. That might include cars hidden behind obstacles, pedestrians emerging from cover or empty parking spaces nearby.
According to the patent application, the shared information would be limited to an area of roughly 50 to 75 metres around the car. That would keep data volumes down and reduce the load on vehicle computer systems.
A patent is not a production feature
GM is not yet talking about a feature ready for showroom cars. For now, this is a patent application. Bringing such a system to market would require fast data connections, common data exchange standards and very clear rules on what information cars may share.
That last point is not a footnote for GM. In 2025, the US Federal Trade Commission accused GM and OnStar of sharing drivers’ location and driving behaviour data without sufficient consent. Under the settlement, the company was banned for five years from passing such data to consumer reporting companies.
GM’s idea of collective perception may well be technically promising, then, but its success will depend on trust as much as algorithms. Cars may one day see round corners for each other. The driver still needs to know exactly who can see the data, and what they plan to do with it.
