Machine learning algorithm for a neural network car- following model

Thesis topic
Driver behavior modeling plays important roles in traffic simulation modeling for transport planning and management as well as development of advanced driving assistance systems. At the Trafik och Logistik (ToL) KTH, we collected driver behavior data via an instrumented vehicle on Swedish roads. Therefore, it is possible to model driver behavior based on the data collected.
This thesis focuses on a particular type of driver behavior, car following, describing driver reactive actions when driving a car behind a leading vehicle. The main task is to develop an adaptive filtering based machine learning algorithm for neural network car-following models using collected data.

Qualification
Students with good analytical background in computer science, electrical and system engineering and being familiar with MATLAB or other computing language are solicited.

Examination
The work will satisfy the requirement of a full-time master thesis in KTH, equivalent to 6 month coursework of 30 credits. The thesis can be examined in several fields such as computer science, electrical engineering, vehicle engineering and transportation systems.

Scholarship
There is possibility to apply research assistant (RA) scholarship so that you can produce a good thesis and get paid.

Contact
Xiaoliang Ma
Traffic and Logistics KTH, visiting address: Teknikringen 72, CTR, Email: [email protected]; Tel: 7908426