Description of field of research
Simulation has been well-established for robotics education and integrated robot software testing for years, there is an ongoing debate in the research community about the ability to transfer robotics skills learned in simulation to reality, a concept termed Sim2Real transfer [1].
Testing learning methods in real robots is often dangerous, since those methods often attempt dangerous behaviours that can damage the robot. For this reason, most successful learning methods for robotics applications are mainly demonstrated in simulation environments. A simulated environment though can never replicate the real world, and methods attempting to transfer learned behaviours from simulation to reality will have to deal with, what is called, the reality gap. The reality gap refers to the discrepancies between the simulated and the real, physical world. The Sim2Real transfer concerns the task of transferring robot skills acquired in a simulated environment to a physical setting [2].
School
Research Area
Robotics | Machine learning | Artificial intelligence
- Research Environment
- Expected Outcomes
- Reference Material/Links
- Our Team
Robotics lab at CSE. J17 room 510.
At the end of the project, it is expected the student will develop and implement algorithms to test state-of-art sim2 real methods in robotics in different simulated and real robot platforms available at UNSW, such as Nao, Baxter, or ARI robot.
- Höfer, S., Bekris, K., Handa, A., Gamboa, J. C., Golemo, F., Mozifian, M., ... & White, M. (2020). Perspectives on sim2real transfer for robotics: A summary of the r: Ss 2020 workshop. arXiv preprint arXiv:2012.03806.
- Dimitropoulos, K., Hatzilygeroudis, I., & Chatzilygeroudis, K. (2022). A brief survey of Sim2Real methods for robot learning. Advances in Service and Industrial Robotics: RAAD 2022, 133-140.
