Abstract
Abstract
Nowadays, road vehicles have more and more autonomous driving capabilities. To perform advanced autonomous driving functions, the vehicles relie on a local description of the environment. This local description is in fact an environmental models that can be used to plan behavior. Since a single sensor may not provide such model, data from several sensors are generally combined using data fusion methods. Indeed, if considering Intelligent Transport Systems related problems, there are sensors in vehicles but also on the road infrastructure. It is thus mandatory to perform Vehicle-to-Vehicle and Vehicle-to-Infrastructure data exchange based on wireless communication to permit cooperation. An Hardware/Software framework called Collaborative Perception Platform (CPF) was designed to share information. Since there are latency times along the chain that induces spatial error on the model, it is of importance to timestamp each information in the same reference timescale for all entities so that the effect of latency times may be partly mitigated using prediction. In this paper, authors present an hardware framework in the context of the Collaborative Perception approach. It consists of a special gateway called SensorHub. It handles data acquisition from sensors or bus and provides these data to a remote unit through an Ethernet link. The SensorHub has onboard timestamping capabilities based on a clock synchronized with the UTC time (Universal Time Coordinated) provided by a GPS (Global Positioning System) receiver. Timestamping functionality and data acquisition are implemented in a FPGA (Field Programmable Gate Array). With such approach, parallel hardware timestamping is performed. Thorough, timestamping functionality assessment were carried out using a reference signal and several sensors.