Connected and Autonomous Vehicles (CAVs) take advantage of the advancement of
communication and sensing technologies to offer a potential sustainable alternative to current
mobility services. Many ongoing projects are studying the effects of CAVs on the network, at
the same time trying to identify the best strategies to develop in order to design new and
dedicated traffic control strategies. The challenge here is to solve the medium-term situation
where both conventional and automated traffic will share the road network.
Very first real-world deployments have been carried out experimentally and tend to confirm
the expected benefits of cooperation between AVs (Stern et al., 2018), that have been
previously identified in simulation (Guériau et al., 2016). However, the high cost of running
such full scale tests (with relatively small amount of vehicles) is one of the reason CAV
technologies deployment takes more time than expected. Indeed, autonomous driving
algorithms, especially when built using Artificial Intelligence techniques and learning-based
approaches, require substantial quantities of data and experience that cannot be solely
provided by actual field-tests.
The deep-learning community is being more and more aware that machine learning models
are “costly to train and develop, both financially, due to the cost of hardware and electricity or
cloud compute time, and environmentally, due to the carbon footprint required to fuel modern
tensorprocessing hardware”1. This is led to the emergence of an interesting and novel research
effort, called Green AI in the litterature (Schwartz et al.,2019), that aims at reducing the amount
of data required for deep-learning based approaches to converge. One the the strategies
“which has been recently gaining importance to drastically reduce computational time and
energy consumed is to exploit the availability of different information sources”1, or
different models, environments.
Simulation and robotic models (such as in Hyldmar et al., 2019) appear to be the fastest
way to first develop, train and then test autonomous driving tasks, allowing the system
designer to investigate the behaviour of embedded navigation systems in a wide range of
situations and/or conditions and across different environments (including for instance
adverse weather conditions that affect vision-based perception, Blin et al. 2020). This process
however faces several challenges: reality gap, when a simulation/model fails to capture all
the particularities of a real system, and domain adaptation, when a model is
developped/trained in a particular context and has to adapt to a different one.
In this context, techniques akin to transfer learning (TL) have been developed to enable
knowledge acquired by one or multiple (Taylor et al., 2019) learner (source) agents to be
transferred to another or the same (target) system, helping the latter to learn a similar but
different task or to adapt an existing algorithm to a similar domain. Recently, TL was shown to
be particularly efficient when transferring autonomous driving tasks, between different
domains (Sharma et al., 2019) and from simulation to a real system (Balaji et al., 2019).
1 https://news.mit.edu/2020/artificial-intelligence-ai-carbon-footprint-0423
The work intended within this Post-doc position is part of a research project named
MultiTrans, that focuses on exploring novel TL approaches for autonomous vehicles scene
semantic segmentation and detection accross 3 different environments (as depicted in the
Figure above: simulation, a robotic platform and a real-world autonomous shuttle test-bed).
This position is funded by the Agence Nationale de la Recherche (ANR) under grant reference
ANR-21-CE23-0032.
Objectives
The main objective is to build a novel autonomous vehicle virtualization framework
(digital twin) enabling to investigate and propose new algorithms that rely on transfer
learning techniques.
More specifically, the work is expected to contribute to the following objectives within
MultiTrans project:
- Identification of critical applications requiring multi-domain transfer (extracted from
real-world domains);
- Scripting of base use cases, small variations (taking advantage of computer power in
simulation), major variations (that require realistic alterations of sensing capabilities)
and corner cases (that could be modelled and tested in the robotic testbed);
- Development of a virtualization framework allowing simulation environments and real
environments to benefit from each other.
Expected contributions and research outreach
The work undertaken by the successful candidate should contribute and is not limited to:
- a better understanding of issues related to implementing autonomous driving across different domains: reality gap, overfitting, few-shots learning, experimental biases, etc;
- insights on the use of simulation and robotics environments to foster and accelerate the development and deployment of connected and autonomous driving technologies.
- novel approaches to transfering and acquiring knowledge from simulation to realworld
and from real-world to simulation;
Given that part of the research will be led jointly using a robotic environment and the newly
developped digital twin, it is expected that the framework developped will allow for reproducible experiments that could be used as demonstrators for research/teaching but also for disseminating material (such as video recordings) to the public.
References
R.E. Stern, S. Cui, M.L. Delle Monache, R. Bhadani, M. Bunting, M. Churchill, N. Hamilton, H.
Pohlmann, F. Wu, B. Piccoli and B. Seibold, “Dissipation of stop-and-go waves via control of
autonomous vehicles: Field experiments”, Transportation Research Part C: Emerging
Technologies, 2018, 89, pp.205-221.
M. Guériau, R. Billot, N.E. El Faouzi, J. Monteil, F. Armetta, S. Hassas, “How to assess the benefits
of connected vehicles? A simulation framework for the design of cooperative traffic
management strategies”, Transportation research part C: emerging technologies, 2016, 67, pp.
266-279.
R. Schwartz, J. Dodge, N. A. Smith, O. Etzioni. “Green ai.” arXiv preprint arXiv:1907.10597, 2019.
N. Hyldmar, Y. He and A. Prorok, "A Fleet of Miniature Cars for Experiments in Cooperative
Driving," 2019 International Conference on Robotics and Automation (ICRA), 2019, pp. 3238-
3244.
R. Blin, S. Ainouz, S. Canu and F. Meriaudeau, “A New Multimodal RGB and Polarimetric Image
Dataset for Road Scenes Analysis”, IEEE/CVF Conference on Computer Vision and Pattern
Recognition Workshops, 2020, pp. 216-217.
A. Taylor, I. Dusparic, M. Guériau, S. Clarke, “Parallel transfer learning in multi-agent systems:
What, when and how to transfer?”, International Joint Conference on Neural Networks (IJCNN),
2019, pp. 1-8.
S. Sharma, J.E. Ball, B. Tang, D.W. Carruth, M. Doude, M.A. Islam, ”Semantic segmentation with
transfer learning for off-road autonomous driving”, Sensors, 2019, 19(11), pp. 2577.
B. Balaji, S. Mallya, S. Genc, S. Gupta, L. Dirac, V. Khare, G. Roy et al., “Deepracer: Autonomous
racing platform for experimentation with sim2real reinforcement learning”, IEEE International
Conference on Robotics and Automation (ICRA), 2020, pp. 2746-2754.
Keywords
Simulation, robotics, transfer learning, autonomous driving, computer vision, deep
(reinforcement) learning.
Qualification and skills
The successful candidate would:
- have completed a PhD. in Computer Science or Robotics with a specialization/interest
in Robotics, Simulation, AI- and/or machine learning-based techniques;
- have demonstrated research experience and relevant publication records;
- have strong English and/or French writing and oral communication skills.
Knowledge and/or experience with the following fields would be greatly appreciated:
- robotics environments (ROS, etc.) and/or vehicular simulation (CARLA, SUMO,etc.);
- intelligent transportation systems, connected and automated vehicles;
- AI techniques such as deep learning, reinforcement learning, transfer learning, multiagent systems.
Supervision
Maxime Guériau (Assistant Professor) and Samia Ainouz (Professor), both member of the
Intelligent Transportation Systems team (STI) at LITIS lab (Laboratoire d'Informatique, de
Traitement de l'Information et des Systèmes), INSA Rouen Normandy, France.
About LITIS lab and the STI team
The research conducted at LITIS lab covers 3 major fields: information access, bio-medical
information processing and ambient intelligence with applications in health, automotive and
smart territories. The expertise of LITIS members is recognized internationally and includes:
machine learning, multi-agent systems, intelligent vehicles. The STI team (the successful
candidate will be joining) is specialized in advanded driving assistance systems, computer
vision, distributed and autonomous systems.
The LITIS is a laboratory (EA 4108) of University of Rouen Normandy, University of Havre
Normandy and INSA Rouen Normandy. It is a member of the doctoral school MIIS and of the
normand network «Digital Normandy». LITIS is a partner of the Normastic CNRS Research
Federation.
Address: 685 Avenue de l'Université, 76800 Saint-Étienne-du-Rouvray.
The candidate will be allowed to access to different experimental platforms to carry out the
work:
- a robotic platform featuring different robot cars equipped with state-of-the-art perception sensors;
- an autonomous shuttle test-bed with equipped infrastructure;
- an intensive computing center (CRIANN: Centre Régional Informatique et d'Applications Numériques de Normandie).
Salary, starting date, travel and support
This 2-year (24-month) Post-doc position is funded by the Agence Nationale de la Recherche
(ANR) under grant reference ANR-21-CE23-0032.
Salary: 2300€ net/month
Expected starting date: around Sept. 2022.
The successful candidate will receive support for occasional international travel (participation
to conferences). Occasional national travel (mainly to Nice and Paris) will be organised,
enabling the candidate to visit the partners of MultiTrans project.
Dissemination and visibility of the work will be ensured by advertising it on MultiTrans
website and sharing resources, open-source algorithms, framework and findings (on a Git
repository).
Application process
Candidates applications should include:
- a full resume, including a comprehensive list of publications, and;
- a cover letter, and;
- contact details of up to 2 references;
And be sent to:
maxime.gueriau@insa-rouen.fr,
samia.ainouz@insa-rouen.fr ,
By no later than May 27 th 2022 .