A multi-level cooperative perception scheme for autonomous vehicles

This paper introduces a multi-level cooperative scheme for autonomous vehicles, using the sensors equipped on-board and a communication scheme with the deployed infrastructures. The proposed model focuses on the communication elements, delving into the cooperative aspect between the network components. The first level of the proposed scheme is composed of a multi-sensor data fusion framework using the measurements obtained from the vehicle on-board sensors, in order to detect obstacles in front of the vehicle. The obstacle detection is based on a liner regression fusion rule, which combines the obtained features from several sensors. Once the decision is taken, the vehicle responds consequently using implicit coordination and bringing the network to an uncertain status. Hence, the second level consists on achieving a stable network by applying explicit coordination between the vehicles and the deployed infrastructures. The third level controls the entire network from a centralized perspective using long-range communication links. Using the knowledge obtained from the centralized network, optimization can be achieved using a coordination scheme based on communication. To finalize, the theoretical framework is simulated under realistic conditions obtaining promising results, in terms of obstacle avoidance and network coordination.