Thesis: Prediction and planning using deep learning

Background
The demand for automated driver systems (ADS) for heavy duty trucks is driven by prospects of increased safety and productivity. On public roads, ADS requires paying attention to the surrounding traffic to enable a frictionless cooperation. As an example, consider doing an automated lane change with a long truck-trailer combination in dense traffic involving multiple surrounding manually driven vehicles. To simplify, surrounding vehicles fall in two categories: those that are affected by the truck’s motion and those that do not need to pay attention since they are in a position where there is no bilateral interaction. The reciprocal interactions in this cooperative game setting is governed by cues of different types, such as blinking and relative positioning. In this example, the truck needs to create a gap between any two vehicles in the next lane by showing intentions to change lanes. In this process, understanding and continuously predicting the motion of other vehicles, as well as planning and showing intentions are important for efficiency and avoiding safety hazards.
Objective and method
The aim of this project is to improve and evaluate traffic prediction and its integration with decision making and trajectory generation for use in automated driving systems on public roads. An apprenticeship learning algorithm capable of predicting multi-modal output will be used for the co-generation. Apprentice learning, a form of supervised learning, is the process of learning from observing demonstrations made by experts. The experts can in this case be common and typical car and truck drivers including their vehicles. It takes as input object motion data and handmade features. Training data consists of two types: simulated vehicles and logged real traffic. An important part of the project is to suggest an appropriate evaluation method of the use-case.
Qualifications
  • Knowledge of deep neural networks
  • Skills in programming in Python, PyTorch/Tensorflow, Linux etc.
  • Master level student primarily from Data science, AI, Computer science, Systems control and mechatronics, Automotive engineering or from other relevant education on master level
Miscellaneous
Thesis Level: Master
Language: English
Starting date: January 2020
Number of students: 1-2
Location: Gothenburg
Tutor: Martin Sanfridson, 031-3220665, Vehicle Automation, Volvo Technology

About Us

The Volvo Group is one of the world’s leading manufacturers of trucks, buses, construction equipment and marine and industrial engines under the leading brands Volvo, Renault Trucks, Mack, UD Trucks, Eicher, SDLG, Terex Trucks, Prevost, Nova Bus, UD Bus and Volvo Penta.

Volvo Group Trucks Technology provides Volvo Group Trucks and Business Area's with state-of-the-art research, cutting-edge engineering, product planning and purchasing services, as well as aftermarket product support. With Volvo Group Trucks Technology you will be part of a global and diverse team of highly skilled professionals who work with passion, trust each other and embrace change to stay ahead. We make our customers win.

We want to get to know you

Application Process

Apply

An email confirmation will be sent as soon as you submit your application. After this, it is still possible to update your personal profile.

Interview

If selected for an interview, you will be contacted with information about the following process steps: second interview, assessments and references.

Selection

All candidates will be notified when a final candidate is selected for the job. You can choose for us to keep your resume in our database and activate a search agent that will look for other jobs that match your profile.

Introduction

When you begin your employment, you will receive an introduction to help you quickly become part of the team and start working with your tasks in the best possible way.

Testimonials

Similar Jobs