Distributed and augmented vehicle perception to support autonomous driving

Principal investigators: Franck Davoine, Véronique Cherfaoui (Heudiasyc)

DAPAD Website





Advanced driver assistance systems have entered the market, with successes like adaptive cruise control and lane keeping assistance. However, a lot of research and development efforts are still necessary to achieve autonomous driving, which involves very complex tasks (such as overtaking and lane changing on motorways, or crossing intersections in urban areas). Robust perception for scene understanding and navigation is still an open problem for which research is necessary. Emerging solutions combine on-board exteroceptive sensors, and are capable of understanding some of the near vehicle’s surrounding environment with static or moving objects, obstacles, traffic signs, the navigable space, etc. But significant progress is still required to augment the range of perception systems that remains limited because of the use of on-board sensors with a field of view restricted to tens of meters and that can suffer from occultation due to other traffic participants or road geometry for instance. The perception capability of each vehicle can be enhanced by wireless information exchange if vehicles in the neighborhood transmit pertinent information on their position, speed and intended actions. This will make it possible to reach a higher level of reliability, consistent with the constraint of driving control and trajectory planning. New approaches need to make a wider use of existing communication channels, vehicular ad hoc networks (VANETs) and other near-future vehicle2x networks to exchange descriptors of the scene they perceive. The project aims at enhancing the vehicle’s perception and situational awareness of the complex and highly dynamic traffic scene, for the sake of a better autonomy, while making use of more sources of information than the one provided in a standalone way by its on-board sensors. Other traffic participants like pedestrians, motorbikes, buses or trucks, all seen as nodes of a mobile ad hoc network, can helpfully share such information.

Figure 1 Collaborative perception: sharing of local sensing information via wireless communications with other vehicles or infrastructures, The perception range can be extended up to the boundary of connected vehicles. The overall uncertainty about the scene can be reduced.
  • Stéphane BONNET
  • Philippe BONNIFAIT
  • Véronique CHERFAOUI
  • Franck DAVOINE
  • Thierry DENOEUX
  • Bertrand DUCOURTHIAL
  • Reine TALJ
  • Philippe XU

Post-doctorals positions:

In the process of recruitment, 2017-2018 (with Véronique Cherfaoui and Franck Davoine): Collaborative perception for autonomous vehicle driving.


PhD students:

  • Romain Guyard, 2017-2019 (with Bertrand Ducourthial and Véronique Cherfaoui): Distributed algorithms for cooperative vehicle perception.

DAPAD’s aim is to design perception methods for autonomous vehicles. In this context, the thesis focuses on the cooperation between vehicles in such a way the embedded sources of information contribute to the estimation of the environment. Tools and techniques have to be designed to process and fuse data coming from all these distributed sources. This is a challenge because sources may send erroneous data either voluntary (e.g. attacks) or involuntary (e.g. unreliable sensors), communication may fail (messages losses...) and the convergence of the distributed algorithms may be delayed due to the network dynamic.

  • Hafida Mouhagir, 2015-2017 (with Reine Talj, Véronique Cherfaoui and PSA Group): Trajectory Planning for Autonomous Vehicles Using Evidential Grids.

We consider the path planning problem for autonomous vehicles given uncertain knowledge about the surrounding environment. We use evidential occupancy grids to deal with sensor uncertainties. We develop a planning approach based on clothoid tentacles allowing a vehicle to move autonomously and safely in an environment which is not perfectly known a priori and in which moving obstacles are present.

  • Elwan Hery, 2016-2018 (with Philippe Xu and Philippe Bonnifait): Localization relative to high definition maps for autonomous and cooperative vehicles navigation.

This is made possible by using a high-definition map containing accurate information, such as lane markings which can be used to reduce significantly the cross-track and heading errors. In the context of cooperative localization, a vehicle can use its perception capabilities along with V2V communication to retrieve the relative pose of another vehicle in order to improve the accuracy of its own pose estimate. This work introduces different methods to compute a curvilinear pose, defined as a curvilinear abscissa, a signed lateral distance and an orientation relative to a path attached to a high definition map.

  •  Hermes Pimenta de Moraes Junior, 2016-2018 (with Bertrand Ducourthial): Data Sharing in Vehicular Networks.

Most of the ITS applications dedicated to vehicular networks rely on periodic messages sent in the vicinity of the vehicles. To ensure the road safety, the inter-messages delay admits strong constraints. The current standard proposes to adapt the inter-messages delay according to the vehicle dynamics. Nevertheless, when the density of vehicles is large, short delays may lead to collisions and losses, leading to a poor neighborhood knowledge accuracy. Recently, we proposed adaptive strategies to take into account the networking conditions for updating the inter-messages delay. The aim is to detect the neighbors in time while preserving the network resources. We show that our cooperative approach achieves very good results in neighbor discovery while consuming less messages.

  • In the process of recruitment, 2017-2019 (with Véronique Cherfaoui and Franck Davoine): Collaborative perception for autonomous vehicle driving.

The objective of the project is to contribute to the definition, development and evaluation in real road driving situations of a collaborative perception system for autonomous and communicating vehicles: a vehicle locally perceives the scene with its own sensors and combines its perception with perceptual or intentional information received from other road users and infrastructure. The thesis work will aim to increase the vehicle's perception range and quality, while estimating the confidence it can have in it. Tests and experiments will be carried out during the thesis through various scenarios involving two to four instrumented vehicles.


Master 2 Internships, Spring 2017

  • Ahmed Ali Hadid (Véronique Cherfaoui and Hafida Mouhagir): Evidential occupation grids;
  • Federico Camarda (Véronique Cherfaoui and Franck Davoine): Fusion of Occupancy Grids in a network of vehicles;
  • Stefano Masi (Philippe Xu and Philippe Bonnifait): Precise vehicle localization;
  • Anthony Welte (Philippe Xu and Philippe Bonnifait): Precise point positioning and protection level;
  • Sergio Pertierre Do Monte (Franck Davoine and Ahmet Sekercioglu): Evaluation of neuromorphic event-based image sensors for dynamic scene analysis.


Master 2 Internships, Spring 2016

  • Elwan Hery, 2015-2016 (with Philippe Xu and Philippe Bonnifait): Cooperative localization for autonomous connected cars.


Engineer student projects, Fall 2015 – Spring 2017

For these projects, Heudiasyc team provides electric vehicles and equipment, partially of entirely funded by the EQUIPEX Robotex project (N° ANR-10-EQPX- 44-01):

  • Two electric vehicles Renault ZOE with their CAN bus. The automated control of the vehicles is done through MicroAutoBox prototyping hardwares programmed by Renault.
  •  Localization equipment: Novatel SPAN-CPT: GNSS + INS receiver with enclosed IMU; Septentrio PolaRx: GNSS reveiver; u-blox 8 / u-blox M8 GNSS receiver.
  • Perception equipment:  4-layer scanners LiDAR Sick LD-MRS; 16-layer 360° scanners LiDAR Velodyne VLP-16; Ibeo Laser Scanners; Intelligent cameras MobilEye; Smartmicro Radar: front radar.
  •  Communication equipment: Cohda Wireless modems.
  •  Test platforms:

The SEVILLE test site is a closed test road inside the Labex MS2T campus domain. It is a closed road composed of two roundabouts with traffic signalizations and lane markings;

The VIL (Vehicle-In-the-Loop) test bed, a semi-simulated environment offering complementary testing capabilities to the vehicles. The road, the obstacles, the infrastructure as well as the perception sensors are simulated within the SCANeRTM studio environment. The longitudinal control of the vehicle (acceleration and braking) is done directly on the testing real vehicle. Four actuators are connected to the wheels to simulate road contact.


Spring 2017: ROS-based vehicle platooning.

Students: Jules Aubry, Sander Ricou, Benjamin Bar, Mewen Michel, Marie Simatic.

Development of a new platoon system using the Robotics middleware ROS (Robot Operating System). ROS is built from the ground up to encourage collaborative robotics software development, internationally. Integration of our sensors and new perception and localization modules in ROS.


Fall 2016: PACPUS-based connected autonomous vehicles.

Students: Benjamin Alix, Guillaume Bizet, Geoffrey Guyot, Miguel Medina, Rémi Sanchez.

Development of a platoon system using the middleware PACPUS. The platoon consists of two vehicles that travel closely together. The lead vehicle controls the speed and direction, and the following vehicle follows the lead vehicle’s movement. PACPUS is a multi-platform (Windows, X86-Linux, ARM-Linux) framework written in C++ in Heudiasyc Lab., that aims at helping the development of applications for Intelligent Vehicles and Autonomous Driving.


Spring 2016: Grand Cooperative Driving Challenge 2016.

Students: Alexandre Coden, Nick Laurenson, Antoine Le comte, John Oudart, Dorian Resmann.

Preparation of Heudiasyc team and participation in the GCDC. Our participation required to develop solutions for vehicle localization, mapping, perception, control and vehicle-to-vehicle communication. Several complex cooperative scenarios were considered, including the merging of two lanes and cooperation at intersections.


Fall 2015: Connected autonomous vehicles.

Students: Florent Thévenet, Quentin Marmouge, SiSi Zhang, Wilhem Devaugerme, Thomas Kieffer, Bastien Fremondiere, Thibault Brocheton, Yann Droniou.

Preparation of Heudiasyc team participation in the Grand Cooperative Driving Challenge (GCDC) 2016 whose aim to boost the development of cooperative automated vehicles by means of wireless communication. The challenge included two scenarios, cooperative platoon merge and cooperative intersection passing, in Helmond, the Netherlands. It involved ten European teams.

On-road data acquisitions / datasets:

Definition of driving scenarios inside, and outside the university campus around the mid-size city of Compiegne, involving up to three vehicles. For this purpose, we equip our vehicles with video cameras, intelligent cameras (Mobileye), 2D/3D LiDAR sensors, GPS, IMUs, and V2x communication tools. 

Accurate ground truth is provided by a Velodyne laser scanner and a GPS localization system.


  • Participation with nine other European teams in the International Grand Cooperative Driving Challenge 2016, in Helmond, The Netherlands;
  • Development of solutions and algorithms for localization and perception systems for autonomous vehicles;
  • On-road data acquisition, experimentations and evaluation of cooperative autonomous driving;
  • Involvement in the main priorities of the newly created joint research unit named SIVALab between Renault and Heudiasyc laboratory. This scientific and technological partnership has been set up in March 2017 for an initial, extendable period of four years. It will use the Renault ZOE-based autonomous vehicle platforms developed by Heudiasyc.


Ph. Xu, G. Dherbomez, E. Héry, A. Abidli and Ph. Bonnifait. System Architecture of a Driverless Electric Car in the Grand Cooperative Driving Challenge. IEEE Intelligent Transportation Systems Magazine, accepted for publication, 2017.

J.-B. Bordes, F. Davoine, Ph. Xu and T. Denœux. Evidential Grammars: A Compositional Approach for Scene Understanding. Application to Multimodal Street Data. Applied Soft Computing, Elsevier, under press, 2017.

K. Lassoued, Ph. Bonnifait, I. Fantoni: Cooperative Localization with Reliable Confidence Domains Between Vehicles Sharing GNSS Pseudoranges Errors with No Base Station. IEEE Intelligent Transportation Systems Magazine, 9(1): 22-34, 2017.

H. Mouhagir, R. Talj, V. Cherfaoui, F. Aioun, and F. Guillemard. Trajectory planning for autonomous vehicle in uncertain environment using evidential grid. The 20th World Congress of the International Federation of Automatic Control, Toulouse, France, 2017.

E. Hery, S. Masi, Ph. Xu and Ph. Bonnifait. Map-based Curvilinear Coordinates for Autonomous Vehicles, IEEE International Conference on Intelligent Transportation Systems, Yokohama, Japan, Oct. 16-20, 2017.

E. Hery, Ph. Xu and Ph. Bonnifait. Along-track Localization for Cooperative Autonomous Vehicles IEEE Intelligent Vehicles Symposium. Redondo Beach, CA, United States, June, 2017.

E. Héry, Ph. Xu and Ph. Bonnifait. One-Dimensional Cooperative Localization for Vehicles Equipped with Mono-Frequency GNSS Receivers. In European Navigation Conference, Lausanne, Switzerland, May 9-12, 2017.

C. Zinoune, Ph. Bonnifait, J. Ibanez Guzman: Sequential FDIA for Autonomous Integrity Monitoring of Navigation Maps on Board Vehicles. IEEE Trans. Intelligent Transportation Systems 17(1): 143-155, 2016.

Ph. Xu, F. Davoine, J.-B. Bordes, H. Zhao and T. Denœux. Multimodal Information Fusion for Urban Scene Understanding. Machine Vision and Applications, Springer, Vol. 27, Issue 3, pages 331–349, April 2016.

Z. Tao, Ph. Bonnifait: Sequential Data Fusion of GNSS Pseudoranges and Dopplers with Map-Based Vision Systems. IEEE Transactions on Intelligent Vehicles 1(3): 254-265, 2016.

C.-L. Yu, V. Cherfaoui, P. Bonnifait: Semantic evidential lane grids with prior maps for autonomous navigation. IEEE ITSC: 1875-1881, 2016.

B. Ducourthial, A. Wade. Dynamic p-graphs for capturing the dynamics of distributed system, Ad Hoc Networks (Elsevier), Volume 50, Pages 13-22, 1 November 2016.

J. Radak, B. Ducourthial, V. Cherfaoui, S. Bonnet. Detecting road events using distributed data fusion: experimental evaluation for the icy roads case. IEEE Intelligent Transportation Systems Transactions, Vol. 17, No. 1, pp. 184-194, 2016.

H. Pimenta de Moraes Júnior, B. Ducourthial. Adaptive inter-messages delay in vehicular networks. 12th IEEE International Conference on Wireless and Mobile Computing, New York, October 2016.

B. Ducourthial, V. Cherfaoui. Experiments with Self-Stabilizing Distributed Data Fusion. 35th IEEE Symposium on Reliable Distributed Systems, Budapest, Hungary, September 2016.

H. Mouhagir,  R. Talj, V. Cherfaoui, F. Aioun and F. Guillemard. Integrating safety distances with trajectory planning by modifying the occupancy grid for autonomous vehicle navigation. 19th IEEE International Conference on Intelligent Transportation Systems Conference, Rio de Janeiro, Brazil, pp. 1114-1119, November, 2016.

H. Mouhagir,  R. Talj, V. Cherfaoui, F. Guillemard and F. Aioun. A Markov Decision Process-based approach for trajectory planning with clothoid tentacles. IEEE Intelligent Vehicles Symposium, Göteborg, Sweden, pp. 1254-1259, June, 2016.