How to use GPS as primary source of data in RTAB-MAP ?

[mrrabbit]

Hi Matlabbe,
I am working on navigating and SLAM for a mobile robot using GPS as localization method. But the problem is, it is failing at some cases due to signal loss at some point in the environment. So I am looking for a SLAM method that does use the GPS as primary source and switched to other slam methods when the GPS goes out of signal and comes back to GPS when the GPS comes back alive. Have any of you guys got any idea about any slam technologies doing this. I tried using RTAB-MAP, but the problem is it uses a combination of all sensors available to it, it does not give priority to GPS as needed. It fuses all these sensor data. Is there anyway to do this.

Thanks for your time.

[matlabbe]

Using a GPS alone may not be enough for localization, is it fused with an IMU? What are the other sensors available?

RTAB-Map uses a Graph-based SLAM approach, in which GPS values are added to the graph as constraints. It needs a least a continuous source of local localization estimation like odometry, then when GPS is not available, it just continues to estimate without those constraints.

Another approach is to fuse odometry with GPS using an EKF filter, like robot_localization package.

In both cases, you need an odometry approach that is able to estimate localization while GPS is not available.