We are looking to run mid-to-long term projects so you can really get to grips with some machine learning. These will be fantastic opportunities to learn how to write and get experience with AI software outside your course.

Have your own project idea that you want to get off the ground?

If so, please send an email to fb534 at cam dot ac dot uk with an outline of the project and we will get back to you as soon as we can!

Projects currently recruiting:

Xilinx Open Hardware 2020 competition

Jon Chuang is currently looking for 2 additional team members, who are undergraduate or PhD students at the University of Cambridge, for the above competition. If interested, please contact Jon at

Find more information about the project here.

Past projects:

Learning to grasp objects with an anthropomorphic passive hand and reinforcement learning

Project Summary
Investigate learning strategies to enable grasping of objects using a passive ‘human-like’ hand mounted on an industrial arm.

Project Description
The human hand has been optimized over generations to enable everyday tasks. The main goal of the project is to investigate how important is the hand morphology in enabling manipulation tasks like grasping.

The students will be provided with an anthropomorphic hand (shown above) attached to a UR-5 platform. The motion of the hand and the object will be tracked with a motion capture system.

The aim of the project is to:

  • Investigate reinforcement learning algorithms that can enable grasping of objects by moving the UR-5 arm and the passive dynamics of the hand. Imitation learning can be used to narrow the search space.
  • To study the effect of the hand dynamics (characterized only by the compliance of the joints for now) on the learning process.

The project results will be published in a conference or journal publication.

Run by Dr. Thomas George Thuruthel

If interested, come to the first meeting on 31 Oct 2019 6pm at the Dyson Centre, Department of Engineering.

Contact Jakub ( for additional details.

Music Transcription with Deep Neural Networks

Starting off with solo piano pieces, develop a machine learning solution that automatically transcribes a music recording into a score.

The project would aim to improve on work done by Google’s Magenta team .

Team leader’s email: