Mounting orientation of MTF-02 Optical Flow

Hello,

Does anyone know the coordinate system for an MTF optical flow sensor such that I can match it with the drone’s coordinates to get the correct flow direction/position estimate?

I understand furniture assembly without looking into manual, but this… it is directly on micoair page: MTF-02P - Optical&Range Sensor - MicoAir Tech

With info that this can be changed in their configuration software if required.

Hey, thanks. As you can see, this info wasn’t available for the MTF-02 pages: MTF-02 - Optical&Range Sensor - MicoAir Tech

My bad. Not “P” version.

Anyway - I am pretty sure OF calibration takes care about orientation as long as it is 90deg aligned, but I am not 100% sure.

But the flow sensor is the same. ‘P’ just denotes that the rangefinder has 4m extra range. So 2m - 2.5m for MTF-02 and 6m for MTF-02P.

Now, the orientation given here seems to hold as I can do flowhold well. But when I manually test the sensor, I get the x and y orientations interchanged (as read from Optical Flow mavlink messages) and so when I modified the flow_yaw parameter to -90 degrees, the drone wan’t holding position and was wanting to fly toward one direction. I also tried it with 90 degrees with the same results.

So I guess this might be because the x and y of the sensor is not the same as that of arducopter?

I tried MTF-02 and this is very optimistic. Usually well below 2m. MTF-02P is way better.

I guess it requires good lighting, but anyway for my purposes, it’s fine

For position and orientation you must define FLOW_xx’s parameters. You can move it in some orientation and check on Mission Planner if movement is the same. Then you can try indoor missions.

Yes I did do this as I have said in my other comment. But it turns out in order for it to work, FLOW_X corresponds to vehicles Y and vice-versa

this is mine

FLOW_ADDR,0
FLOW_FXSCALER,0
FLOW_FYSCALER,0
FLOW_ORIENT_YAW,0
FLOW_POS_X,0
FLOW_POS_Y,0
FLOW_POS_Z,0
FLOW_TYPE,5

Edit: add footage

and this is when she hovered indoor with loiter

I think the final best test is an indoors mission: waypoints can be virtual on a well known place. Then you define the EKF origin on the home waypoint.

Indoor missions with optical flow are possible on a 3mx3m or less space. However, I experience drift, so for a second mission I define again the EKF origin.