View job on Handshake

Motion Planning Intern

As part of Motionals Motion Planning Research team you will create and implement new cutting-edge machine learning methods that push the boundaries of the state-of-art in order to bring reinforcement and behavior cloning based methods for controlling driverless cars to the streets!

You will be an integral part of a team of highly talented and motivated researchers within Motional that are given the freedom to work on cutting edge long-term projects. The ideal candidate would therefore share the team’s vision of trying new experimental approaches while not being afraid to fail and possess the enthusiasm to never stop trying until a break-through is achieved.

What You’ll Be Doing

Current MPC formulations use hand-crafted cost functions and constraints in order to generate trajectories. The availability of differentiable MPPI (Model Predictive Path Integral Control) as well as large collections of human driving data enable us to train a neural network that describes the cost function and the implied constraints. Given a human driven trajectory we will select the applicable homotopy which together with the ego state will be the input to the cost function network. This enables us to employ a data driven approach to trajectory generation that can exploit the known model dynamics.

  • You will not have to start from zero as there has already been a lot of impressive work on this project by other members of the team.
  • Your expertise in large-scale machine learning projects and MPPI will help your team achieve break-throughs in getting this project to fruition. You will contribute directly to our code-base and if successful you will be able to deploy your code for on-road testing on our Hyundai Ioniq 5 driverless fleet.

What We’re Looking For

  • You’re passionate about a driverless future and how machine learning will play a crucial role in saving more than a million lives every year and enabling equitable mobility for all.
  • You have a strong track-record of getting machine learning solutions to work.
  • You are able to take responsibility and push forward projects without close supervision.
  • You have a deep understanding of machine learning techniques and you are able to find solutions for problems that are not handled by literature.
  • You are proficient in pyTorch and Python.
  • You have experience in large-scale training of machine learning based methods.

Bonus Points (not required)

  • Experience in large software projects
  • Solid control theory knowledge
  • You are proficient in C++
  • You have implemented MPPI-based methods by yourself or worked in a project that used it.

Why You Should Join Us

  • Because you can change the world with us.
  • We have a positive culture where experimentation is encouraged and you will be given the freedom to do things your way.
  • You will have the chance to work with some of the brightest people in the industry.
  • We have autonomous cars on the road.