THESIS WORK - DETECTION OF ROAD ANOMALIES USING SUPERVISED LEARNING ALGORITHMS

Thesis Worker at Volvo Cars Welcome to explore the world of Volvo Cars by writing your thesis with us! As a thesis worker in our organization you are supported by a supervisor who follows you during your project.
All thesis projects are arranged in business critical areas and therefore you will be able to contribute to our company purpose – providing freedom to move in a safe, sustainable and personal way – from day one!    Title.
Detection of Road Anomalies using Supervised Learning Algorithms     Background.
With the enhanced computational power available for Autonomous Driving (AD) and Advanced Driver Assistance Systems (ADAS), supervised learning algorithms have become feasible for road condition prediction applications.
Additionally, future cars will come equipped with advanced perception stacks, including cameras, radars, ultrasonic sensors, and lidars.
This enables next-generation cars to fuse the sensor information and execute complex algorithms in onboard CPUs and GPUs.
Previously, motion-based methods were developed using vehicle and suspension motion signals.
In this thesis, the focus is to evaluate the detection of road anomalies using supervised learning on perception sensors, thereby facilitating further enhancement of the road safety of Volvo Cars.
Scope.
The thesis work should study different algorithms that can detect anomalies and classify the road ahead of the vehicle, according to ISO standards.
The algorithm should demonstrate the ability to extract at least one of the following road surface anomalies.
•    Bumps •    Potholes •    Cracks •    Damages •    Deep manhole covers •    Road roughness •    Road repair patches  •    Road vertical profile classification which are interesting for AD/ADAS warning and path control functions, as well as active suspension adjustment.
Supervised learning algorithms that detect these road features will start by using only camera data and then, if required, other perception sensor data, e.
., LiDAR.
The feasibility of the algorithm will be evaluated using a subset of thousands of kilometres of offline proprietary data available at Volvo Car AB.
The scope could also extend to test runs of the algorithm on Nvidia hardware in a test vehicle.
The thesis work will include the following parts.
•    Literature review of state-of-the-art road anomaly detection and road classification methods.
•    Literature review on camera-based supervised learning algorithms.
•    Ground truth data will be provided for the road anomalies.
•    Estimation accuracy and uncertainty analysis of sensor data against ground truth data.
•    Investigate suitable supervised learning algorithms, ranging from support vector machines to deep neural networks, convolutional neural networks and                    transformers neural networks.
•    Selection of appropriate performance metrics for evaluating and comparing proposed learning algorithms.
•    Open-loop simulation and verification of the proposed algorithms.
•    Stretched target.
Stretched target.
Implementation of the designed algorithm on Nvidia hardware connected to sensors in a test vehicle.
The study will last 20 weeks (30 ECTS, MSc thesis).
The work will be carried out at Volvo Car AB.
Suitable Student background  MSc programs within Data science/AI, Embedded systems or related fields.
Knowledge in Mechatronics/Robotics, Vehicle Dynamics, Computer Vision is a merit.
Experience with Python or C++ is a must.
Starting date.
January     Number of students.
2 Location.
Göteborg, Volvo Car Corporation, PV2A, Volvo Jakobs väg Göteborg.
Category/Subcategory.
Vehicle State & Lateral Control   How to learn more Attach your resume, courses and grades, and cover letter stating your interests within the given area and your thoughts and credentials.
If you want more information about the project or simply learn a bit more about the team, please reach out to.
Tutor / Contact person    Derong Yang, VCC  Arun Vijayan, VCC,  Anton Larsson, VCC,   Volvo Cars - Driving change together​ Volvo Cars’ success is the outcome of a collaborative, diverse, and inclusive working environment.
Today, we are one of the most well-known and respected car brands, with around employees across the globe.
At Volvo Cars, your career is designed around your skills and aspirations, allowing you to reach your fullest potential.​ By , we aim to sell 1.
million cars annually, with 50 per cent being electric cars and sold directly to customers mainly through digital channels.
Make sure you’re in the front seat on this exciting journey as we pioneer the driving and electrification technologies of the future.