For this quarter, I will work with Tsangkai addressing on EM-SLAM problem. First half of quarter:
- Investigate EM-SLAM problem in detail. Validate the result from "EM-SLAM with Visual/Inertial Application" using Python.
- Try to achieve convergence of this algorithm. If unable to achieve, figure out the reason why it fails and try potential methods
- Give a demo of robot in 2D case and analyze the performances.
Second half of quarter:
- Convergence with EM-SLAM is usually not guaranteed. I will focus on developing new ways to make this algorithm converge.
- EM-SLAM is an offline algorithm. I will try to develop an online version of EM-SLAM algorithm and give a demo