Handling Occlusions in Automated Driving Using a Multiaccess Edge Computing Server-Based Environment Model From Infrastructure Sensors

Abstract

The on-board sensors’ view of an automated vehicle (AV) can suffer from occlusions by other traffic participants, buildings, or vegetation, especially in urban areas. However, knowledge of possible other traffic participants in the occluded areas is crucial for an energy and comfort optimizing control of an AV. In such a case, information from infrastructure sensors sent via vehicle to anything (V2X) communication can help the AV. Fur such cases, we have developed and prototypically implemented a concept where an infrastructure environment model is generated from infrastructure sensors on a multi-access edge computing (MEC) server of an LTE/5G mobile network. This information extends the AVs’ field of view and is beneficially integrated into their motion planning schemes. In this article, after a description of the modules of our approach, we present and discuss real-world results from our pilot site on a public junction with prototype AVs.