**Current Projects:
Deep Reinforcement Learning for Robust and Sample Efficient Large Scale Robotics Platforms
The project consists in developing learning algorithms to achieve robust and efficient large-scale logistic solutions such as bin picking, conveyor sorting, loading and unloading in real-world environments.
In particular, we are interested in leveraging invariant and equivariant principles to learn to multiply a robot’s experience throughout the state-action space across algorithms and settings. We are also interested in better understanding causality and invariant factors to lead to better learning algorithms.
In doing so, we hope to tackle outstanding problems in DRL including sample complexity, generating large and meaningful datasets, and better generalizing. DRL has yielded very useful policies and robust behavior in logistics and other areas (i.e. QT-OPT), however, at a very high cost in training.
If, DRL is to be widely adopted in industry it must be able to learn more efficiently.
We are looking for people with strong backgrounds in programming, machine learning, and deep learning or candidates who have shown very promising potential to join us in advancing the state-of-the-art in DRL. We are looking for Research Assistants, Post-Doctoral Fellow’s, Ph.Ds, , Master, and Undergraduate students. Are all welcome.
Please reach out to me for further details.
Research
Thank you for taking the time and consideration of looking into my group and research. I have take the time to write a message to prospective students. I may at times receive a large number of inquiries from prospective students and may not be able to respond personally to each one. Please accept this response.
Graduate Students and Research Assistants
My research delves in how we can learn to manipulate in more intelligent ways. We are researching robust deep (reinforcement) learning algorithms that can be applied to a wide-range of logistic scenarios. We are interested in developing algorithms that can learn faster, better, and from less experience.
Well-rounded students with exceptional ability in programming, probability and inference, artificial intelligence, machine learning (supervised, unsupervised, deep learning), and reinforcement learning as well as a great deal of passion and curiosity in learning and solving problems.
You will be a full-time user of Linux, ROS, Torch and/or Pytorch, and simulators. You will use advanced editors like vim, emacs, and usually write in LaTex in Overleaf.
Working in our lab and at CUHK will allow you to do cooperative work all across the world: from North America, to Europe, and throughout Asia.
I appreciate an inquisitive mind, great team work spirit, and a very strong initiative to get things done. The interested student should have a good awareness of the latest scientific contributions in the fields related to the lab. I expect students to build their ‘idea stack’. That is, 2-3 research hypothesis. These ideas will be your constant guide in the midst of waves of information, papers, and new paradigms.
Students with a strong publication record or implemented project are attractive if you have them. Interesting publication venues in the area of robotics are listed here.
In order to be part of my team, we will have to negotiate an important and interesting fundamental research question that you will spend your time investigating during the period of your graduate studies. At the same time, you should seek to: (1) master a body of background literature related to the robotic fields relevant to our group; (2) build a strong technical background in robotics that supports independent and innovative research (3) master experimental tools and methods of empirical analysis; and (4) the development of communication skills for the clear and effective presentation of your results orally and in written form.
Prospective Post-Doctoral Positions
There are immediate openings for post-doctoral fellow and research assistants/associates in dexterous manipulation for logistic tasks focused on learning/reinforcement learning at the CUHK T Stone Robotics Institute [http://ri.cuhk.edu.hk/] (CURI) at the Chinese University of Hong Kong (CUHK).
CURI:
The CURI institute is a multidisciplinary institute integrated by around 30 faculty members spanning AI, robotics, and medicine. The institute has a strong history of pioneering research as well as successful technology transfer. In addition, to the lab space on the main campus, CURI enjoys ample laboratories in the CUHK Shenzhen Research Institute and the Futian (Shenzhen) Innovation Research Institute. Also note that the Chinese University of Hong Kong ranked the 43th in the world by the latest QS ranking.
HKCLR:
Furthermore, CURI began a vibrant collaboration with the newly established Hong Kong Center for Robotics Logistics (HKCLR). A not-for-profit center that works in tech transfer and opens up incredible opportunities to transfer research to industries in Hong Kong, China, and beyond. All in all, CURI enjoys a very strong and large multi-disciplinary team spanning all areas of robotics are working together to make significant research and industry contributions.
Collaboration:
In addition, there is a new and robust collaboration initiative between the University of California, Berkeley, and the HKCLR center. Close partnership, students and staff exchange, and annual symposiums will be taking place.
Research:
Research projects and development concern advancing state-of-the-art algorithms in dexterous manipulation and deep reinforcement learning applied to logistic scenarios and working towards advancements in sample efficiency, robustness, safety, and learning efficiency as well as failure identification, classification, and recovery.
Postdoc Qualifications:
– Ph.D. in Computer Science, Mechanical, Electrical Engineering or related field.
- Strong publication track record in top journals and conferences in robotics and learning.
- Strong hands-on experience in deploying deep (reinforcement) learning and imitation learning algorithms in real robots. Preferably deployments with robot manipulators and large-scale tasks.
- A passion for advancing learning.
Benefits:
Monthly salary and fringe benefits: salary will be highly competitive, commensurate with qualifications and experience. For postdoctoral fellow and research assistant, contract is typically yearly based and renewal is upon mutual agreement.
Contact:
Interested candidates should also send their complete CV, top publications, and project portfolio to Dr. Juan Rojas at juan [dot] rojas [@] cuhk [dot] edu
Prospective PhD Students
PhD candidates are encouraged to apply via the CUHK Graduate School application, or through Hong Kong’s Research Grants Council (RGC) Ph.D. fellowship scheme application. The fellowship endows a PhD students with around 43,000USD per year as well as nearly 2,000USD in conference travel allowance per year.
Currently I co-supervise Ph.D. students with Prof. Liu Yun-hui. Final decisions of supervision are done by him.
Prospective Research Assistants/Associates
There are immediate open positions for RAs to work on deep reinforcement learning in dexterous manipulation. Pay is commensurate with experience and competitive internationally.
RAs will help in deployed production systems and will gain great experience in working with SOTA methods. It can be a great springboard for future PhD positions.
Qualifications for Research Assistant/Associates:
– A Master’s or Ph.D. degree in in Computer Science, Mechanical, Electrical Engineering or related field.
– Promise of strong performance via solid project demonstration or a top paper.
Contact:
Interested candidates should also send their complete CV, top publications, and project portfolio to Dr. Juan Rojas at juan [dot] rojas [@] cuhk [dot] edu
Prospective Master Students and Undergraduate Students
We welcome master and undergraduate student’s to join deeply into the research of the group. Over the years we have had a large number of master students and some undergraduate students publish one or various high quality papers to top journals and conferences. There is certainly opportunities and space to do great research for motivated students.
General Advice
In this section, would like to share general advice that I have found helpful over time.
On Research…
Many of these come from Bill Freeman from CSAIL MIT
How to do Graduate-level Research:
Some Advice by Dr. Bhaskar Krishnamachari
How to do Research
by Bill Freeman from CSAIL MIT
How to have a Successful Research Career
by Bill Freeman from CSAIL MIT
On Publication…
How to write a good CVPR Submission
by Bill Freeman from CSAIL MIT
On Machine Learning…
Advice on Machine Learning Projects
by Tim Rocktaschel
On Careers in Academia…
Check this study based on surveys of post-docs