UROP Openings

Have a UROP opening you would like to submit?

Please fill out the form.

Submit your UROP opening

UROP: Object Detection and Tracking with Point Cloud using Deep Learning for Roboat


Term:

Fall

Department:

11: Urban Studies and Planning

Faculty Supervisor:

Fabio Duarte

Faculty email:

fduarte@mit.edu

Apply by:

09/31/2020

Contact:

Tixiao Shan: shant@mit.edu

Project Description

Roboat is the world’s first major research program on autonomous floating vessels in metropolitan areas. It allows for creating dynamic infrastructure, transportation of goods and people, and environmental sensing on Amsterdam’s canals. While the first prototypes of self-driving cars are taking to the road, Amsterdam ushers in a new chapter in the international push for autonomous vehicles. Roboat is the result of innovative large-scale research that explores and tests the rich possibilities for autonomous systems on water. It represents a fantastic opportunity and privilege for the city of Amsterdam to have the world’s most prominent scientists work on solutions with autonomous boats - especially in a location where water and technology have been linked for ages. Roboat is a joint research program of Massachusetts Institute of Technology (MIT) and Amsterdam Institute for Advanced Metropolitan Solutions (AMS Institute). The City of Amsterdam, Amsterdam’s water company Waternet and the City of Boston are supportive of the 5-year program. The novel robotic boat is easy to manufacture, highly maneuverable, and capable of accurate trajectory tracking in both indoor and outdoor environments. The project allows for innovations in various domains, stands strong with top research and knowledge expertise, and receives univocal support. Areas of innovation are for instance assembly, sensing, and waste removal. The principal investigator and professor of MIT Carlo Ratti highlighted the dynamic and temporary floating infrastructure such as bridges or stages that can assemble on-demand in a matter of hours. Opportunity To enhance the navigation capabilities of Roboat, we are incorporating cutting-edge technologies from the autonomous driving industry to our autonomous surface vehicle. More specifically, we are looking to utilize point cloud-based object tracking technologies that enable our boats to interact with the canals of Amsterdam more intelligently. The goal of object tracking is to identify and track objects that belong to one or more categories in the canal, such as boats, bridges, animals, and other objects. We expect object tracking task to play an important role in our project for obstacle avoidance, action recognition, behavior analysis, and so on. Evaluation The outcome of the student will be evaluated on real-hardware in real-world environments. We will test the system designed by the student on a real boat in the canals of Amsterdam. Other details One position will be open for the proposal Student wage: $13/hour (maximum 140 hours) The project covers necessary travel and material costs The position is remotely available

Pre-requisites

Background in computer vision, robotics, or related field Be able to learn or have experience with deep learning tools, such as PyTorch, Tensorflow, or Keras Be able to learn or have experience in programming using Python Be able to learn or have experience with Robot Operating System (ROS) (optional)