Relocation of an ADF in Learning mode not working?

问题

I have a strange behaviour when trying to append to an existing ADF:

I'm loading an ADF which was just recorded and the device can easy relocate on. Once I load the same ADF with learning mode on (in order to extend the existing ADF) the device cannot relocate on it. It's easy to reproduce (see the link to the video): - Record an ADF - Load it, make sure the device can re-locate - Load it again with learning mode "on", the device can no longer re-locate on it

I tried the explorer-app the java area-learning sample as well as the unity area learning sample.

In my own Application I do check the status of the re-location with the following snippet in "onPoseAvailable":

if (pose.baseFrame == TangoPoseData.COORDINATE_FRAME_AREA_DESCRIPTION
            && pose.targetFrame == TangoPoseData.COORDINATE_FRAME_START_OF_SERVICE) {
        currentrelocationStatus = pose.statusCode == TangoPoseData.POSE_VALID;

Any ideas what is going on here? Or did I miss something and it's supposed to work in a different way?

I'm running the latest release and I'm building with the Kajam Libs.

Thanks in advance for any help...

Video-Demo

回答1:

On current Tango software SDK, it is known that extending ADF will make the relocalization much harder. You have to walk around much longer in order to get it relocalized. Under the layer, extending ADF is running another pipeline other than only relocalizing with an ADF.

回答2:

I have the same problem (my stackoverflow question). I'm also working with Java and the Mira release.

@Jason Guo: Do you want to say, that I have to do localization as a first step and then after that I should extend my ADF?

来源：https://stackoverflow.com/questions/36882384/relocation-of-an-adf-in-learning-mode-not-working