Lidar Calibration for Autonomous Vehicles

Lidar is a key sensor modality for autonomous vehicles, providing a detailed 3D representation of the vehicle surroundings. This representation is provided as input to obstacle perception and localization, the functions of the autonomous system providing the vehicle with an understanding of what is around it and where it is, respectively.


In most cases, to achieve an optimal view of the surroundings, multiple lidar sensors are required and are mounted at well-chosen locations on the autonomous vehicle. For data from multiple sensors to be usable for obstacle perception and localization, a precise understanding of the rigid relationship, or transforms, between sensors relative each other and between sensors relative the vehicle origin is needed. The process for determining these transforms is known as the calibration of the sensor set.


Lidar sensor calibration can be determined by measuring in a CAD model of the vehicle, with the drawback that variations in tolerances and mounting precisions are not accounted for. It is also the case that, due to the often extreme vibrations encountered by the vehicles in operation, sensors can move slightly on their mountings. Since lidar is observing the world at a range of up to 100m, small sources of error can have a significant impact on the performance of the autonomous system.
Thesis scope and milestones:


The scope of this thesis is to investigate methods for accurately determining the calibration of a lidar sensor set using the lidar data and any of the available on-board sensor modalities, such as IMU, GNSS, Odometry, etc.


Milestones:
Literature review of existing lidar calibration approaches.
Selection and implementation of the most promising methods.
Validation using simulated data
Validation using logged data



Who are you?


We are looking for two curious, engaged and highly motivated students comfortable with programming and with strong mathematical skills. This thesis project will include various fields such as lidar point cloud processing and registration, optimization, and simulation.


Who are we?
The Environment team is part of the global technology organization at Volvo Autonomous Solutions. We design and implement the perception, localization, and mapping software components of the autonomous system, as well as handle all questions related to sensors and sensor set design.
https://www.youtube.com/watch?v=T-TOCGC6Uuc


What is our offer to you?


Imagine being surrounded everyday by a diverse, fun, and passionate group of colleagues every day who you get to work with and enjoy eating lunch and having fika with. Now imagine that you work with these colleagues on some of the most advanced autonomous vehicle technology in the world. Add that you are working with these colleagues on this advanced technology in a culture that promotes growth, collaboration, trust and is striving to be the best autonomous solution verification and validation organization in the world!


Are you ready to join our team as a thesis worker and be part of shaping the future of the transportation industry with us?


Thesis Level: Master
Language: English
Starting date: January 2023
Number of students: 2
Tutors:
Anders Sunegård, SW and Algorithm Developer Localization, +46 739022801
Tomas Halleröd, SW and Algorithm Developer Localization, +46 739025422


Curious, and would like to know more? Contact one of the tutors or
Ian MacIsaac, Manager Localization & Mapping, +46 739025861
https://youtu.be/aUNiVxmiJDE