Application Deadline : June 30,2021
Huawei is working on key components of L2-L3 autonomous driving platform and progressively shifting focus to development of breakthrough technologies required for L4-L5 autonomy. Tomorrow self-driving cars powered by AI will combine edge and cloud compute with vast number of sensors to safely and autonomously drive customers and deliver merchandise. At Huawei we develop a full-stack of technologies to enable this dream, including compute units, sensors, communication and cloud. We are seeking the best candidates for PhD CIFRE with a background in computer vision, deep learning, reinforcement learning, mapping, perception, sensor fusion, cognition and other related areas, to work as a part of IoV team in Paris Research Center (PRC). As a member IoV PRC you will closely work with multiple teams worldwide to grow your expertise and successfully transfer your research results into real products.
The ImViA Laboratory (Imaging and Artificial Vision) is a research unit that groups the activities in Computer Vision of the University of Burgundy. This laboratory, created in 2019, is structured in three groups IFTIM (Medical Imaging), CoReS (Real Time Computer Vision), ViBot (Vision for Robotics). The PhD will be directed by Cédric Demonceaux (ViBot) and Cyrille Mignot (CoReS).
The core subject of this PhD should focus on improving onboard mapping for long-term localization in a dynamic environment by leveraging crowd-sourced data. Localization and mapping are fundamentals component to enable safe self-driving and advanced automation capabilities for cars. However, accurate mapping is costly in term of data temporal resolution (in order to keep the map up-to-date, data need to be collected often) and data quality (multiple sensors with different modalities). An appealing solution consists of using crowd-sourced data with high temporal resolution but lower quality compare to data obtain from an expensive dedicated mapping platform [1,2]. Leveraging crowd-sourced data enable to capture the dynamic of the environment in order to update the map representation used by localization algorithms of the car. The temporal analysis of these data open the possibility to predict changes willing to occur as well as distinguish between permanent changes and temporary structural modification, in a data-driven manner . In order to share crowd-sourced information between entities, data acquired by an embedded system should be compressed or pre-processed onboard. It involves using light mapping algorithms that can be run on low-computational capability devices and applying filters designed to retain significant map features for targeted applications, such as localization . Map quality qualification is also an important aspect for a localization service provider, such as the one used in an autonomous car. By analyzing the quality and the density of the crowd-sourced data, a self-diagnostic can offer a fine qualification of the mapped area. Such qualification permits to proactively request new data on specific zone to maintain a desired level of map quality needed to safely operate an autonomous vehicle.
 Schonberger, Johannes L., and Jan-Michael Frahm. “Structure-from-motion revisited.” Proceedings of the IEEE conference on Computer Vision and Pattern Recognition (CVPR). 2016.
 Platinsky, Lukas, Szabados, Michal, Hlasek, Filip, et al. “Collaborative Augmented Reality on Smartphones via Life-long City-scale Maps”. Proceedings of the IEEE International Symposium on Mixed and Augmented Reality (ISMAR). 2020.
 Zi Jian Yew and Gim Hee Lee. “City-scale Scene Change Detection using Point Clouds”. Proceedings of the IEEE International Conference on Robotics and Automation (ICRA). 2021
 Arthur Moreau, Nathan Piasco, Dzmitry Tsishkou, Bogdan Stanciulescu and Arnaud de La Fortelle. “CoordiNet: uncertainty-aware pose regressor for reliable vehicle localization.” 2021.
Description of research activities:
- Study state of the art on crowd-sourced mapping, online mapping, change detection, map qualification, maintenance and compression
- Identify bottleneck in map maintainability and promising research direction toward dynamic map representation for long-term localization
- Propose new solutions for unsupervised map update based on crowd-sourced data
- Research and develop algorithm based on the proposed solutions
- Apply proposed algorithm to the domain of self-driving cars using existing or specifically collected datasets
- Publish research results in top conferences and participate to scientific seminars
This PhD will be supervised jointly between Huawei Technologies France and the laboratory ImViA of the University of Burgundy.
The candidate should be motivated to carry out world class research and should have a Master in Computer Vision and/or Robotics. He/She should have solid skills in the following domains:
- Implement Code in Python & C++
- Apply or use existing libraries for deep learning in project related tasks
- Good knowledge in Git, ROS, OpenCV, Boost, multi-threading, CMake, Make and Linux systems
- Code and algorithm documentation
- Project reporting and planning
- Writing of scientific publications and participation in conferences
- Fluency in spoken and written English; French and/or Chinese is a plus
- Intercultural and coordination skills, hands-on and can-do attitude
- Interpersonal skills, team spirit and independent working style
Cédric Demonceaux (EMR VIBOT CNRS 6000, ImViA) – email@example.com
Cyrillie Migniot (ImViA) – firstname.lastname@example.org
Nathan Piasco (Huawei) – email@example.com
Dzmitry Tsishkou (Huawei) – firstname.lastname@example.org
CV + motivation letter + transcript of records for academic years 2019-2020 and 2020-2021 + any other relevant document