Week 18 19

WEEK 18-19

Self-localization with SLAM technique

• I have been studying two SLAM based algorithms, MonoSLAM and PTAM. I have decided to move towards PTAM because of the advantages it introduces over the other one.
  • PTAM allows relocation
  • Manage more points
  • The location is more precise

• First I am developing a solution using a real camera, and then I will take it to the simulated exercise.

• References:
  • https://github.com/eugeniu1994/S-PTAM-Stereo-Parallel-Tracking-and-Mapping
  • https://github.com/luigifreda/pyslam
  • https://github.com/uoip/stereo_ptam
• To Do:
  • Finalize the algorithm to run in the simulator

MonteCarlo Visual Location

• We decided to use an algorithm that uses the ceiling and a single pixel of the image to self-localize.
• A roomba model with the camera facing the ceiling has been added.

• A ceiling model is being designed that has enough textures for the robot to be able to locate itself.

• To Do:
  • Finalizing the roof model