Particle filter to track robots
Particle filter demo of our home-made robot, the dotty. The IP webcam used in this demo is quite old, so the quality is not that good. You will see that the pockets of the pool table are quite hard (the particle filter is tempted to track those instead of the robot when the robot gets close). You can find the code for the particle filter at github, feel free to improve it.
There is not much time spend on this implementation. It is only meant as an exercise to get familiar with particle filtering techniques. There are approximate and generalised forms of belief propagation on networks of robots that require such expertise. Subsequently this can be used for either simultaneous localisation and mapping, tracking, or certain common visual tasks.
With respect to the latter, it would be cool if an image segmentation task across a multi-agent system would indeed be solved as one general problem where the structure is exploited, rather than some ad-hoc user-defined decomposition depending on the coffee quantity in the morning.
For other algorithms do also not hesitate to visit the website NETS dedicated to algorithms only!
blog comments powered by Disqus