Master Thesis - Outlier detection for Volvo Trucks Safety Scoring
Volvo Business Services AB / Maskiningenjörsjobb / Göteborg
Observera att sista ansökningsdag har passerat.
Visa alla maskiningenjörsjobb i Göteborg,
Mölndal,
Partille,
Kungälv,
Lerum eller i
hela Sverige Visa alla jobb hos Volvo Business Services AB i Göteborg,
Mölndal,
Kungsbacka,
Borås,
Trollhättan eller i
hela Sverige One of the objectives stated on Volvo 's group mission is to be 100% safe. In this context, evaluating drivers' behavior according to safety plays a vital role. Human made mistakes like over speeding, inattention, fatigue are important factors for accident causation and must be dealt with to reach the above-mentioned objective.
Outlier detection is a well-established research field with several applications ranging from fraud detections for credit cards, health care and network security. A big set of truck data from several fleets can be analyzed and unsafe drivers can be viewed as outliers, identified, and trained to improve their safety performance.
Problem
Today Volvo Trucks has a service called Safety Reports that provides information to fleet managers regarding safety behavior of the drivers in their fleets. The service also includes a Safety Score that ranks the drivers based on their driving style.
The current score logic was developed using basic histogram analysis and it can identify only big deviations and safety behavior and no traffic context.
Solution
By introducing more advanced outlier detections it would be possible to identify more nuanced deviations of safety behavior and more precise traffic context that could explain possible low scores.
Goal of the thesis
Investigate possibility of using advance outlier detection on Safety Driving assessment and traffic context assessment. Appling the investigated techniques to Volvo test fleet and produce a case study. This objective can be divided in the following sub-objectives:
Identify the best date set coming from the vehicle that can be used to assess safety driving and traffic context.
Prepare a literature review on the state-of-the-art of Outlier detection.
Identify the best technique to apply to this context.
Do a case study on a Volvo test fleet to demonstrate the results.
Desirable expertise
Signal and systems
Statistical Analysis
Artificial Intelligence
Linux, Python, C, R
Kick-off date
The thesis can get started immediately.
Additional info
The scope can be flexible and adapted to 1-2 students, depending on how many students you are and how much time you have for your thesis.
Some reference links:
Accident Research Team Safety Report - 2017
Accident Research Team Safety Report - 2022
A critical overview of outlier detection methods
A Review of Local Outlier Factor Algorithms for Outlier Detection in Big Data Streams
Outlier Detection: Methods, Models, and Classification
Så ansöker du Sista dag att ansöka är 2024-01-15
Klicka på denna länk för att göra din ansökan Omfattning Detta är ett heltidsjobb.
Arbetsgivare Volvo Business Services AB (org.nr 556029-5197)
405 08 GÖTEBORG
Arbetsplats Group Trucks Technology
Jobbnummer 8237352
Observera att sista ansökningsdag har passerat.