Collaborative, GPS-free Techniques for Localization in Miniature Robots

Localization within the environment is still a big challenge for robotic systems and especially for minimalist, insect-size robots used in swarm intelligent systems. Odometry is the simplest approach to this problem, and can be realized by keeping track of the speed of each wheel to calculate the robot’s relative displacement. Unfortunately, odometry is highly unreliable due to wheel slip and encoder noise.

In one of his projects, Collaborative, GPS-free Techniques for Localization in Miniature Robots, Nikolaus Correll is exploring a simple collaborative approach to reduce this error inspired by the way that flocking birds manage to return to the same nesting point from half way across the world. Taking simple flocking as a baseline, we will be able to further examine more complex collaborative algorithms.

Moreover, for analyzing collective behavior of mixed animal/robot societies, the SWIS group has developed SwisTrack, a tracking software allowing for tracking societies of a priori known dimension.

Relation to my thesis: Just being curious on the way other fields approach similar challenges on collaboration, space, context and failures/limitations. Collaborative localization system is stolen from the nature. It might have some value for the human world. Plus, they use tracking software to analyze collective behaviors. It has a common ground with the replay tool we use in Catchbob! to analyze group collaboration and location awareness.