Integration of Belloto leg detection implementation

[hawesie]

@ferdianjovan has been integrating @marcobecerrap's implementation of a laser-based leg detection algorithm from Belloto with the aim of getting better long-distance readings from the laser in order to get longer people tracks.

I'm just putting it here so other tracker users know it's going on @cdondrup @marc-hanheide

[marc-hanheide]

where is it? ;-)

[hawesie]

@ferdianjovan ?

[ferdianjovan]

https://github.com/marcobecerrap/leg_detector

[hawesie]

And the configuration for integrating it into the overall tracker?

[ferdianjovan]

You need to follow the instruction in bayes people tracker package https://github.com/ferdianjovan/strands_perception_people/tree/hydro-devel/bayes_people_tracker add the leg_detector to the detectors.yaml. The topic is /leg_detector_topic/.

[cdondrup]

I guess @marc-hanheide is already looking into releasing it? Should we have a stable fork in strands, is Marco going to work on it or is it just maintained from now on?

[marc-hanheide]

I am and talked to @ferdianjovan about it, because the name was clashing (see https://github.com/marcobecerrap/leg_detector/issues/3) and also the package.xml needs fixing: https://github.com/marcobecerrap/leg_detector/issues/4

[ferdianjovan]

I will talk with @marcobecerrap to fix the clashing problem.

[cdondrup]

This seems to be done. Any news on the final integration with the tracker? @marc-hanheide @ferdianjovan I also talked to Nicola and he offered to help with the integration and the filter of false positives using the tracker like it was done in the original paper. No time frame on that, though.

[ferdianjovan]

Not really sure about the integration, but the leg detector is ready in @marcobecerrap repo. That sounds good if there is a filter for all false positives.