Industrial PhD student–Predictive & energy efficient highway autopilot

In this project we will develop optimal controllers for heavy duty vehicles by optimizing energy efficiency and autonomous driving in uncertain environments. This is a challenging problem that includes both long-horizon look-ahead prediction and planning, for optimizing energy efficiency, and a shorter-horizon planning, with main focus on ensuring the safe interaction with other road users. These two objectives need also be realized in real-time, which brings additional challenge on the choice of control algorithms. In this project we will focus on deep learning approaches, and Reinforcement Learning (RL), in particular, but we will also investigate synergies with model predictive control. The lack of interpretability of black-box approaches will be compensated by computing a measure on the level of confidence the controller has when taking a certain decision. We will investigate different approaches to quantify a statistical confidence level. Taking critical decisions will be avoided in situations where the controller is not confident. These same situations may be used to propose methods on deciding the best way of enriching the training set and thus improve the controller confidence.


This project will be part of the Wallenberg AI, Autonomous Systems and Software Program (WASP) funded by the Knut and Alice Wallenberg Foundation. The PhD student will enjoy the benefits of the WASP graduate school https://wasp-sweden.org/graduate-school/, where skills can be acquired on artificial intelligence, autonomous systems and software. As a PhD student you will be given opportunities for many inspiring conversations, plenty of autonomous work and some travel. Short term visits may be organized to highly ranked universities under the WASP umbrella, such as Stanford University, UC Berkeley, Nanyang Technological University, or similar.


This project will provide knowledge that can be used by the heavy vehicle industry to produce world-leading transport solutions. Volvo Trucks is the main applicant and provides industrial delivery. Academic supervision will be provided at Chalmers University. The PhD student will get an office at Chalmers University, and access to education and facilities.


The industrial PhD period is about 4.5 years.


Experience and knowledge
M.Sc. degree in Applied Physics, Applied Mathematics, Systems and Control, Robotics or similar. Excellent grades are expected in most of the subjects
Preferably more than 3 years of hands-on experience in advanced control systems, robotics, or automotive systems and especially vehicle motion control and dynamics
Fluency in English, both written and spoken
Solid mathematical background used in, e.g., physical modelling with differential equations, numerical optimization, control theory, deep learning, statistical modelling
Optimization or applied mathematics in general
Control theory
Statistics, probability, random processes, or similar
Neural Networks, Reinforcement Learning, or similar
Knowledge in optimal control, or model-based control is meritorious
Knowledge in Python is meritorious
Knowledge with version control is meritorious

Competencies
Has a deep interest to figure out how things work and propose new inventive solutions for how things can be solved in the future.
Collaborates - Building partnerships and working collaboratively with others to meet shared objectives.
Action Oriented - Taking on new opportunities and tough challenges with a sense of urgency, high energy, and enthusiasm.
Problem solving - Making sense of complex, high quantity and sometimes contradictory information to effectively solve problems.
Being resilient - Rebounding from setbacks and adversity when facing difficult situations.
Builds Networks - Effectively building formal and informal relationship networks inside and outside the organization.
Ability to bring ideas and projects from theory into practice, make it work in real vehicles

Personal Attributes
Organized - Structured with attention to details and has the ability to manage several topics simultaneously.
Communicates effectively - Developing and delivering multi-mode communications that convey a clear understanding of the unique needs of different audiences.
Interest in research and science, building knowledge and innovation long-term.
Takes ownership of own area and deliveries and thereby take accountability as well to be a team player and put the group/team first
Can-do attitude and being resilient – Not afraid of trying. Rebounding from setbacks and adversity when facing difficult situations.

For questions, please contact Leo Laine +46 76 5535311.


Kindly note that due to GDPR, we will not accept applications via mail.