Open vkee opened 7 years ago
Hi Vincent,
For relocation, it follows the loop closure approach described in the paper "Appearance-Based Loop Closure Detection for Online Large-Scale and Long-Term Operation". In the code, it is implemented here. These lines in the code match the lines 12-14 of the algorithm 1 of the paper.
The Bayes filter is updated here, and inside that function, the prior is updated here (new hypotheses are based on the previous ones).
cheers, Mathieu
Hi,
I've been trying to figure out the behavior of rtabmap when localizing using an existing map, particularly on the graph back end. I haven't been able to figure it out and was wondering if you could give me a brief explanation and point to where it is happening in the code (i.e. when the first relocalization occurs, is a new prior set? and then what happens for future relocalizations, if anything? additional priors set?)
Thanks! Vincent