|Topic:||Path prediction for driving assistance systems based on machine learning.|
|Supervisor:||Jakub Mareš Garant: Doc. Ing. Karel Zimmermann Ph.D.|
|Description:||In recent years, automotive industry has focused on development of ADAS (Advanced Driving Assistance Systems) with ultimate goal to produce fully autonomous vehicles in near future. Valeo R&D Center Prague is currently developing some of ADAS systems like Autonomous Emergency Braking (AEB), Autonomous Cruise Control (ACC) or some other related systems based on several sensor setups.
Both functions mentioned above require solid prediction of future path of ego vehicle based on driver´s input (steering wheel, throttle, brake pedal, etc.) and basic odometry signals (velocity, acceleration, yaw rate, rotation of wheels, etc.). Such predicted path is then usually used by target selection algorithms in order to analyze possibility of collision with other vehicles or to choose a target vehicle for following during cruise.
|Instruction:||Student´s task will be to explore possible utilization of machine learning techniques on this problem. For starter, gaussian processes seem to be a viable choice. However, student may try to apply some other algorithms - the concrete topic will be settled after a discussion between student and supervisor. Training and testing data will be provided from Valeo´s side.
|Bibliography:|| Carl Edward Rasmussen and Christopher K. I. Williams; Gaussian Processes for Machine Learning; The MIT Press, 2006. ISBN 0-262-18253-X
 Christopher Tay, Christian Laugier. Modelling Smooth Paths Using Gaussian Processes. Proc. of the Int. Conf. on Field and Service Robotics, 2007, Chamonix, France. 2007.
|Realization form:||C++ source code, trained model|