Data Fusion considering ‘Negative’ Information for Cooperative Vehicles

Abstract

Negative information provides important additional knowledge that is not exploited for sensor data fusion tasks by default. This paper presents a new approach to incorporate such information about unoccupied, observed areas or missing measurements in the Kalman filtering process. For this purpose, a combination with a grid-based method is proposed to generate a visibility map. This enables a plausibility check and an enhanced understanding for the collaborative perception of the environment with multiple cognitive vehicles. Results from a realistic traffic simulation are presented.