Master Thesis/Student Assistent: using tactile sensor data
Why are robotic hands so clumsy? In an attempt to solve this problem, we investigate the use of tactile sensors for robotic in-hand manipulation. In particular, based on a number of tactile sensors (the BioTac; the DLR tactile sensor; and the iCub tactile sensor), we want to create algorithms that can use such sensing to manipulate objects. In cleartext: write, unscrew a bulb, or hand over a knife with a robotic hand.
We have previously built a test setup with which tactile data can be measured. You can use this setup but have to extend it to fit your experiments. There are two experiments that have to be done:
- measure curvature from a tactile sensor;
- measure/predict slip when sliding over an object.
We are looking for one student for each task; ideally, the two students can cooperate in adapting and using the test bed. In your thesis, you will have to do both practical work (i.e., work with an existing test setup) and programming (i.e., develop the algorithms for data analysis).
You will be working at TUM (Stammgelände). Knowledge of machine learning methods, e.g. from following our ML1 course, are required. Programming experience in Python or Matlab expected.