30hp- Algorithm for automatic evaluation of smart sensor performance using high performancele

The Sensor Fusion group is responsible for providing an accurate model of the vehicle surroundings for the advanced driver assistance systems (ADAS). The group has an important role in the development of several interesting functions such as Advanced emergency braking, Adaptive cruise control etc. A thesis project at Scania and Sensor Fusion is a great opportunity to work on the forefront of ADAS development and an excellent way of making contacts for your future working life.

Background:
Robust sensor performance is key to ensure that our functions work as expected, so for each new sensor project a lot of effort is spent collecting and analysing real world data from the sensor to make sure that it fulfils our needs and requirements. In these scenarios we lack a reliable ground truth reference and usually have to resort to time consuming manual log analysis to evaluate whether the sensor performance is reasonable. To solve the ground truth problem, we plan on mounting some high performance sensors on our trucks that are good enough to serve as reference sensors, but too expensive to use in the end product. In order to speed up the evaluation we also want to automatically evaluate the performance of the sensor and for this we need a stable and reliable association algorithm that connects the reference detections to the reported data from the sensor under test. Once a reliable association is made the sensor performance can be evaluated with respect to the key metrics of interest. The goal of this thesis is to develop and evaluate such an algorithm.

Objective:
To develop an offline algorithm that uses vehicle data and a high performance reference sensor to automatically measure the performance of some less performant sensor that will be used for the actual ADAS system. Some key metrics of interest is 2D position and velocity of the object, but other features, e.g. size and shape could also be of interest. The algorithm should be evaluated using recorded sensor data from real world traffic and test track. If the algorithm is performant enough it could the be used to automatically detect deviations in the specified sensor performance so that we can quickly file an error report to the supplier or in other ways act on the problem.

Your profile:
We are looking for a master student with a strong mathematical background, preferably with a focus on signal processing, sensor fusion or similar. Experience with tracking algorithms is beneficial. We also see that you are responsible, have good collaboration skills and have an interest in working with agile methods.
Language requirements: English (mandatory) and Swedish (preferable)
 
Number of students: 1
Time: 20 weeks, full time 40 hours per week
Start: Jan 2025
Credits: 30hp
 
Contacts and supervisor:
Adam Wettring, Product owner for Sensor Fusion, 0855372348, adam.wettring@scania.com

Application:
Your application should include CV, cover letter and transcript of records

A background check might be conducted for this position. We are conducting interviews continuously and may close the recruitment earlier than the date specified.