The main difference of the propsed algorithm with the previous ones is that each robot tracks the state of all system, even though it has no information on what is going on.

The actions one robot can take include

  1. Motion propagation
  2. Landmark observation
  3. Robot observation
  4. Communication between robots

In these 4 actions, only the last one involves other robots, while the rest all take place locally. In addition, the first one definitely increases uncertainty to the estimates, while the rest are more complicated.

The main challenge now is to quantify the effect of landmark/robot observation.

The detail can be found in my GitLab repository.

Tsang-Kai

Next Post Previous Post