ShanghaiTech Undergraduates at the MARS Lab

Undergraduate students of ShanghaiTech are welcome to join the MARS Lab at any time. You will be able to work with state of the art hardware on various interesting topics related to robotics. CS students but also EE students are welcome. In special cases we may also host students from one of the other ShanghaiTech schools (e.g. Pyhsics or Life). 

Working in a lab early in you student career to get some research experience is a very good idea. If you do a good job, there is a good chance that you can write a paper with us as a co-author (e.g. see here). 

Procedure:

  • Write Prof. Schwertfeger an email (soerensch @ ) that you are interested. Include your CV (or at least some data, e.g. when you joined ShanghaiTech, which school, etc.). If you know already, also write down what topics you are interested in.
  • You can also (before or after writing to Prof. Schwertfeger) just visit the MARS Lab (SIST 1D-203) and talk to the students. The students will give you a short tour and introduce some of the topics we need help with.
  • Take a few days to decide on a topic, then let us know. Most likely the topic will be supervised by one of the lab's graduate students. 
  • Be sure to sign the safety rules.
  • Ask to be given the "MARS Lab Guidebook", read it completely, as it outlines many important guidlines and rules for the MARS Lab. 
  • After some time, when you did some good work already, you can be added in the peoples page of the MARS Lab. 
  • Once you joined our lab, you are invited to join our weekly group meetings! In Fall 2019 they are Fridays, 10:00 in SIST 1D-202. 

Tips:

  • Be persistent! We will not chase after you. If you are really interested its your job to show up!
  • Ask for meetings with the graduate students or the Prof. if you need guidance or want to show some good results.
  • If you have results, share a video/ photo/ paper about your work with the group in wechat or send an email to the Prof.
  • Prof. Schwertfeger may not answer your email - but usually he will take a look at all emails!
  • If Prof. Schwertfeger doesn't answer your important email, you are very welcome to send another email the next day, and the next - no problem! It shows that you are really interested!
  • At the very beginning you may want to learn ROS. For that you'll need a Linux Laptop (if in doubt, use Ubuntu). 
  • It is also a good idea to take a look at the Robotics course lecture slides to get familiar with some of the basic concepts and terms. 
  • You may select a topic that you came up with yourself, but quite likely you cannot expect us to put much energy on supervising you on that. 

Topics:

  • Topics change frequently. Best is to ask the students.
  • This is an older list, some of the items are still relevant: https://robotics.shanghaitech.edu.cn/research/topics 
  • We are participating in RoboCup Rescue (news: here, here and here). Help is always welcome. If you contribute substantially, we may take you along to a competition.
  • We do research regarding mobile manipulation (see the robots) - you could help there.
  • Another research topic revolves around mapping datasets, mapping algorithms and the mapping robot hardware. (see publications and robots).
  • We also work with cameras. We do research on visual odometry (VO), visual SLAM and related topics. (see publications).
  • There are several other (smaller) research areas going on in the MARS Lab.
  • The topics are selected regarding their application and usefullness for real robotc problems. We typically do not choose a research area or technique just because it is a hot topic recently. So if you come to my lab and say "I want to work with XY" (XY begin, for example, Deel Learning or Reinforcement Learning) I will be all for it - IF you can justify the use of said technique for that application. E.g. for image recognition Deep Learning is state of the art. Using it for problems for which we have good algorithms already (e.g. visual servoing) may be a possible scientific research direction - but if most likely the classical solutions will work best, we will not endorese such a project in the MARS Lab.