Perception coopérative inter-véhicules pour la sécurité routière – PERCOIVE

The PERCOIVE project (PERception COopérative Inter-VEhicules pour la sécurité routière) is a candidate to the ANR 'Jeunes Chercheurs' proposition. This project is presented by young researchers of the CNRS-associated laboratory 'Heudiasyc'. The project is planned for three years with a total cost of about 799,972.00 €. The team is composed by 5 associated professors and 2 research engineers. Other members may join the team, namely 2 master students, 2 post-doctoral for periods of 5 and 12 months each respectively. The aim of PERCOIVE is to increase the vehicle safety bye investigating vehicles-on-board sensors, environment perception and eventual cooperation between vehicles. The existing experimental platform PACPUS (Perception et Assistance pour une Conduite Plus Sûre) of the Heudiasyc laboratory will be used in the project execution. This platform is constituted with human resources, two automobiles, and shelters for real time experimentations. Pragmatically, the project is divided into five related 'Actions' which are listed hereafter. Action 1: This action is concerned with robust multi-modal and multi-model parameter estimation. Developing multi-modal estimation methods will be necessary in order to cope with drawbacks of existing estimation methods in the field of parameter estimation. Multi-model parameter estimation will be adequate in the case of different driving situations (road elevation and lateral inclination, irregularities on the road …). Existing 'mono-models' are in general designed by taking into account specific concerns of the environment and fails when we search for generality. In this action, each vehicle is concerned with proprioceptive on-board sensors only and the 'cooperative' aspect will appear in following actions. Action 2: This action is concerned by the integration of the exteroceptive perception of the environment into the vehicle dynamics estimation methods. The aim will be to understand and develop vehicle dynamics parameter observers which do not consider only the conventional proprioceptive on-board sensors, but as well perception of the environment performed by on-board exteroceptive sensors like cameras, laser scanners, radar, GPS, GIS. Action 3: This action focus on the cooperative parameter estimation. Each vehicle will use information from the neighborhood in order to correct the parameter estimation (dynamics or road attributes). The developed methods should take into account heterogeneous information and eventual conflicts between different sources of information. Action 4: This action is concerned with inter- and intra-vehicle communications and will be developed together with the three actions listed above. The aim is to provide a robust distributed assistance drive system by developing a secure communication system to exchange cooperative information in the neighborhood. Action 5: This action is concerned with the experimental test-bed of the cooperative estimation. The aim is to develop hardware and software and their implementation on experimental platforms in order to evaluate the proposed methodologies.