Deep Learning has seen significant growth and improvements in computer vision systems over the last years. Our work at the Australian Centre for Robotic Vision is looking at how robotic manipulation tasks be improved by using visual feedback. A lot of activities at the intersection at robotics, vision and learning at QUT.
We are investigating data driven approaches to robotic grasping and manipulation. In particular, we aim to develop economic ways to collect data - use COTS robots and cheap cameras - and train this systems - reduce the amount of training data needed.
There has been some preliminary work on the QUT Arm Farm. This projects offers opportunitieas for highly motivated students to be part of a bigger goal through VRES, honors and PhD placements.
The aim is to build an autonmous system that can learn to interact with the world. The full system will involve engineering and research efforts, with varous subsystems being developed currently.
More information at http://Juxi.net/projects/ArmFarm/
Skills and experience
We are looking for motivated students with experience in at least two of the following:
- Robotics (building and deploying)
- Implementing machine meaning algorithms
- Analysis and preparation of data (for learning systems)
- Computers vision (running and developing object detection algorithms)
Contact the supervisor for more information.