Frankly speaking, optical flow generates ranging sensor on itself.
Algorithm developed can turn 2D video into 3D video on-the-fly, extracting Z-dimension from 2D video/ comparing movement of selected points between individual image frames.
This is the solution I have offered @Cala to track problems with her crops
(sunflowers).
Video camera pointed down can replace Lidar lite acting as single-point scanner vs.
2D scanner made of video camera + 2D > 3D video conversion to extract Z-dimension to get 3D scanner, hot item in 3D printing industry.
Lidar is a very accurate sensor for measuring distance, maintaining a specific distance from the ground.
Optical flow is a whole different animal, a down facing camera looks at the ground to determine movement and then processes the movement in to a flight path correction.
Optical flow, when it is fully implemented will allow for a non GPS based position hold.
Please note the following....
Load the correct firmware as per the instructions
Focus the camera as per the instructions (look for a red box explaining this)
Tune the flow sensor to ensure sensor rates, and gyro rates are co-ordinated.
Ensure sensor orientation is set correctly for your frame....
It's been a month or so since I've messed with the px4flow. I was at a very similar place with OF that you are now, Wes. It took me a while before someone over on the Px4 Groups on Google told me that the only sure fire way to confirm the OF Data is through enabling the full log bitmask and viewing the log, using FlightPlot or LogMuncher to view the OF Data.
My first mistakes were; Not having a distance sensor working in sync with the OF sensor (bought a LIDAR-Lite), after that I found out I was using Mission Planner Stable and not the Dev/Beta loaded in the Hawk, then I was told that QGroundControl w/ the PX4 Firmware is the preferred firmware for OF. So that brings me here, I've uploaded logs onto log muncher which didn't help me much more than FlightPlot.(Log from test flight). I'm breaking out multiple SD cards tomorrow and starting over with different firmware's and settings. Hopefully I'll uncover something to contribute.
I think it depends upon who you talk to.. so if you talk with the px4 developers they'll lead you the way that they know (i.e. QGC+px4stack), while Paul and I would steer you towards ardupilot + mission planner. So this thread is really for the ardupilot+MP method that hopefully Paul and I can help you with.
Mounted to an Alien 680mm frame, with a hawk loaded with 3.3 r1, Lidar lite installed and functioning well...
Having followed the Ardupilot wiki for the PX4flow...
PX4flow loaded with the Ardupilot recommended firmware.
Optical Flow enabled in Mission Planner.
PX4flow focused through Q ground.
Connected to the i2c expansion board shared by the GPS compass and remote USB and mode light indicator.
Loging set per wiki to log OF and log while disarmed.
Powering up the multi by connecting the battery, the PX4flow lights flash.
Having got this far I have not been able to get any useful data to be displayed in the OF log.
All I get is a single flat line in the log.
I have tried different start up procedures:
Connect battery, arm safety switch, do tilt cycles, disarm safety switch, unplug battery, connect USB download log.
I have also tried connecting USB first, waiting for hawk to start up, then connect battery, arming safety switch, doing tilt cycles, disarming safety switch, unplugging battery, connecting USB, downloading logs.
Is there a way to verify that the PX4flow is working properly apart from the logs?
A line to look for in the mission Planner Stats?
I have had this PX4flow sensor for more than a year, bought it the first time I saw it on the Hobby King website. Then found out that there was very little support for sensor at that time.
Now I am very eager to get the sensor working, what should I try next?
Replies
What's the difference in flying performance between optical flow and Lidar lite?
Frankly speaking, optical flow generates ranging sensor on itself.
Algorithm developed can turn 2D video into 3D video on-the-fly, extracting Z-dimension from 2D video/ comparing movement of selected points between individual image frames.
This is the solution I have offered @Cala to track problems with her crops
(sunflowers).
Video camera pointed down can replace Lidar lite acting as single-point scanner vs.
2D scanner made of video camera + 2D > 3D video conversion to extract Z-dimension to get 3D scanner, hot item in 3D printing industry.
(no more Laser equipment warning label)
Optical flow is a whole different animal, a down facing camera looks at the ground to determine movement and then processes the movement in to a flight path correction.
Optical flow, when it is fully implemented will allow for a non GPS based position hold.
Thank's Wes.
Please note the following....
Load the correct firmware as per the instructions
Focus the camera as per the instructions (look for a red box explaining this)
Tune the flow sensor to ensure sensor rates, and gyro rates are co-ordinated.
Ensure sensor orientation is set correctly for your frame....
What I am not seeing is any useful data being generated by the sensor in the logs.
Is there a way to see live data to verify that the PX4flow sensor is working correctly?
It's been a month or so since I've messed with the px4flow. I was at a very similar place with OF that you are now, Wes. It took me a while before someone over on the Px4 Groups on Google told me that the only sure fire way to confirm the OF Data is through enabling the full log bitmask and viewing the log, using FlightPlot or LogMuncher to view the OF Data.
My first mistakes were; Not having a distance sensor working in sync with the OF sensor (bought a LIDAR-Lite), after that I found out I was using Mission Planner Stable and not the Dev/Beta loaded in the Hawk, then I was told that QGroundControl w/ the PX4 Firmware is the preferred firmware for OF. So that brings me here, I've uploaded logs onto log muncher which didn't help me much more than FlightPlot.(Log from test flight). I'm breaking out multiple SD cards tomorrow and starting over with different firmware's and settings. Hopefully I'll uncover something to contribute.
@SwitchedMode,
I think it depends upon who you talk to.. so if you talk with the px4 developers they'll lead you the way that they know (i.e. QGC+px4stack), while Paul and I would steer you towards ardupilot + mission planner. So this thread is really for the ardupilot+MP method that hopefully Paul and I can help you with.
Having followed the Ardupilot wiki for the PX4flow...
PX4flow loaded with the Ardupilot recommended firmware.
Optical Flow enabled in Mission Planner.
PX4flow focused through Q ground.
Connected to the i2c expansion board shared by the GPS compass and remote USB and mode light indicator.
Loging set per wiki to log OF and log while disarmed.
Powering up the multi by connecting the battery, the PX4flow lights flash.
Having got this far I have not been able to get any useful data to be displayed in the OF log.
All I get is a single flat line in the log.
I have tried different start up procedures:
Connect battery, arm safety switch, do tilt cycles, disarm safety switch, unplug battery, connect USB download log.
I have also tried connecting USB first, waiting for hawk to start up, then connect battery, arming safety switch, doing tilt cycles, disarming safety switch, unplugging battery, connecting USB, downloading logs.
Is there a way to verify that the PX4flow is working properly apart from the logs?
A line to look for in the mission Planner Stats?
I have had this PX4flow sensor for more than a year, bought it the first time I saw it on the Hobby King website. Then found out that there was very little support for sensor at that time.
Now I am very eager to get the sensor working, what should I try next?
-
10
-
11
-
12
-
13
-
14
of 14 Next