The invention discloses an intelligent vehicle cooperation method based on distributed shared memory. Specific to the adaptive capacity of a single intelligent vehicle, a decomposition and re-synthesis method is adopted, and a uniform behavior decomposition and arbitrator scheduling method is used, so that the development process of the intelligent vehicle is simplified. Moreover, extension is performed in a multi-vehicle cooperation scene by using a distributed shared memory technology. Shared memory exists in each vehicle participating in cooperation as a key element of an intelligent vehicle cooperation application, cooperation variables for describing the cooperation situation are vehicle shielding communication details and are used for ensuring consistence. Vehicles read the cooperation variables through simple read-write operation, and cooperation is realized by applying a distributed algorithm on the basis of these cooperation variables. Through the technology in the invention, the distributed shared memory can be constructed effectively and fused into a single vehicle development process, and very high practicability can be ensured in a typical ideal practical cooperation scene.