Postdoctoral fellowship (two years) in Sim-To-Real Transfer of Robotic Controlled Forest Machines
Department of Physics at Umeå University (www.physics.umu.se) conducts strong research in the areas of organic electronics, condensed matter physics, nanotechnology, photonics and theoretical physics
The Department of Physics is looking for a postdoctoral fellow for a project focusing on sim-to-real transfer of robotic controlled forest machines operating in unstructured environments. You will join the research group in Digital Physics and collaborate with roboticists at the Department of Forest Biomaterials and Technology at Swedish University of Agricultural Sciences. Application deadline is Jan 29, 2021. Starting date according to agreement.
Using artificial intelligence (AI), a forest machine system can learn the ability to grasp, load and sort tree logs using data from many different situations. With deep reinforcement learning, the system gradually becomes "smarter" as it discovers strategies that give higher reward for being fast, energy efficient and robust. Doing most of the training in a simulated environment is necessary for economic, practical and safety reasons. It also has the advantage that the AI can be made well-informed about the physical nature of the environment and the machine by gaining access to observations, unavailable in the real world, from the physics engine driving the simulation. The project shall explore how to make capabilities learned in virtual environment transferable to the physical domain. Other research questions may include how to interface the AI with the machine's control system and to a human operator. The postdoctor is given unique access to well-instrumented forest machines for experiments, high-performance simulation software, digital models, and will collaborate with a multidisciplinary research team and industry partners at the forefront of forestry automation and digitalization.
The project is for two years. The stipend is financed by the Kempe Foundation and amounts to 300 000 SEK per year (the stipend is tax-free) plus a budget for equipment and other expenses.
The successful candidate should hold a PhD degree or a foreign degree that is deemed equivalent to a PhD degree in automation, computer or computational science, robotics or equivalent subject. Priority should be given to candidates who have completed their doctoral degree no more than three years before the application deadline unless special circumstances exist.
The applicant should have experience of and strong interest in doing experiments with robots or other mechatronic systems. The applicant should be highly motivated and have the ability to do research independently as well as a part of the research group. The candidate must have very good skills in written and spoken English.
The application should include:
A motivation letter describing previous research experience relevant to the project (max 2 pages).
Curriculum vitae with a list of publications.
Copy of the PhD thesis.
Copies of two most relevant publications.
Certified copies of doctoral degree certificate and other relevant degree certificates.
A list of university courses with grades.
Contact information of three reference persons.
Other documents wished to be claimed.
Your complete application, marked with reference number Dnr FS 2.1.6-2684-20 should be sent in a pdf-document to email@example.com (state the reference number as subject) on January 29, 2021 at the latest. The application should be written in Swedish or English.