Gael wrote,
 
"I use the inmoov-develop which is what I recommend to builders now, because of the ease of the configurations.

I have modified many times the gestures to launch face tracking and at some points( or MRL version) it sometimes worked and then with further releases my gesture would be out of function.

Here is the link to the current gestures:
https://github.com/MyRobotLab/inmoov/blob/develop/InMoov/gestures/trackHumans.py
https://github.com/MyRobotLab/inmoov/blob/develop/InMoov/gestures/trackPoint.py
https://github.com/MyRobotLab/inmoov/blob/develop/InMoov/gestures/stopTracking.py

 

And face recognizer

It was working great when I modified these gestures, but I haven't tested recently due to OpenCV not setting the filters.:

 
I started a Tracking service with a virtual Arduino - (I'll repair my pan/tilt kit soon).  It appears to work in that all the services started up.  It connected to the virtual arduino, the filters & face detect started, opencv started, all appeared worky so far.
 
It would be helpful if some InMoovers could test this, I'm at a bit of a disadvantage without a "real" inmoov.  It would help also if there was any detail regarding "gestures being out of function".  Did they do nothing? Errors ?  Something Else ?
 
 
Ok .. starting to see some problems.  The Servo gui has changed a lot !  More stuff exposed, which is a good thing. Might be nice to re-arrange some of the items.  When I start, the pin, nor controller is set on the Y axis servo. Also on the X when I enable the servo (seems disabled initially) - it autoDisables, but the autoDisable checkbox does not reflect that.
 
 

dom14

6 years 8 months ago

Thank you Grog, i check when i can do.

For the facerecognizer, do you add face detect coordinate ?

Good work.

hairygael

6 years 7 months ago

hairygael: Unfortunately face tracking is not working properly for me with 2483 with inmoov-master

After lauching voice command "search humans" the head is thrown up and to the left, the filters in OpenCV are correctly activated.

It seems like the PID is inverted, When ever I place my face in front of the robot eyes, the head is trying to look away. It's trying to set my face out of the square of face tracking.

When inverting the servos it doesn't solve the pb.

 

 

 

 

It was working for me (see picture above) .. and I get the same results as you Gael now ...

Not sure what happened, I looked at the calculated value when I was doing a test to center LK tracking - 
The filters are correct, the value calculated was 87 and the center was 90 .. but the servo "ADDS" the 87 producing crazy big location for servo to go to ...

The PidTest is worky - although its a crazy simple test ...

This will need more research -  feels like I fix it and it keeps getting unfixed :(

Must be an evil ghost...