Marine Robotics Group

MIT Computer Science and Artificial Intelligence Lab

Interested UROPs

Depending on the work going on in the lab we may have specific project opportunities for interested UROPs. If you are interested in a UROP, please check the listings below. If there is no listing but you're interested in our work, feel free to reach out to mrg-urops@csail.mit.edu and we may have a suitable project for you.

In general, when reaching out please email a resume, and 1-2 paragraphs on why you want to be involved. While there are no hard requirements, you're likely to get more out of your UROP experience with our group if:

  • You have contributed to a substantial project in the past (e.g., prior work experience or personal projects)
  • You have taken some advanced robotics/computer science classes (6.141 is particularly relevant here)
  • You have prior experience with ROS
  • You have prior experience with C++/Python
  • You are able to commit 15+ hours/week

UROPs can be for credit or for pay through the MIT UROP system. If looking for one of these options we must send an application to the UROP Office before the respective deadline.

Object-Based Robot Navigation

We are interested in one or more UROPs to support our research on object-based representations for navigation. Our existing work deals with the problem of building object-level maps of the world from vision (see here for more information). A UROP project could involve extending this work in one of several ways, including:

  • using these object-based maps to support navigation decisions on a real robot
  • improving the accuracy and robustness of our mapping techniques by incorporating learned object descriptors
  • building more accurate object shape models.
Prior experience with ROS/Python/C++ would be helpful, e.g. 6.141.

Multi-Robot Navigation

One or more UROPs to develop autonomous navigation behaviors for our in-house developed robot platform (see here for more information). Potential UROP projects involve:

  • integrating trajectory/path-following abilities into the platform
  • collecting a large dataset to test the performance of multi-robot SLAM algorithms in a wide range of environments
  • integrating camera/LIDAR robot localization solutions into the platform so the robot can self-localize as it navigates its environment
Previous experience with ROS/Python/C++ is suggested. Primary mentor: Alan Papalia (contact: apapalia at mit dot edu)