Here’s a quick technical post for anyone attempting to harness the capabilities of a Realsense D435 camera on a Jetson TX2. For me, this is about getting usable depth perception on a UAV, but it has proved more problematic than I originally anticipated.
This post aims to provide some simple instructions that now work for me, but took a long time to find out!
The Intel librealsense2 library does not support ARM architectures as I write. This causes a fatal compile error when the file librealsense/src/image.cpp is accessed, as it queries the system architecture.
Modify image.cpp as in my Github gist here. This bypasses the architecture check.
- Flash the TX2 with JetPack 3.2
- git clone https://github.com/IntelRealSense/librealsense.git
- cd librealsense
- sudo apt-get update
- modify librealsense/src/image.cpp as per gist above
- sudo apt-get install git cmake
- sudo apt-get install libssl-dev
- sudo apt-get install libusb-1.0-0-dev pkg-config libgtk-3-dev
- sudo apt-get install libglfw3-dev
- sudo apt-get install libudev-dev
- sudo apt-get install cmake-curses-gui
- sudo cp config/99-realsense-libusb.rules /etc/udev/rules.d/
- sudo udevadm control –reload-rules && udevadm trigger
- mkdir build && mkdir install
- cd build
- cmake ../ -DBUILD_EXAMPLES=true -DCMAKE_BUILD_TYPE=release -DBUILD_UNIT_TESTS=false
- make -j4
- sudo make install
- Connect D435 to the USB3 port
- Navigate to the tools at usr/local/bin, such as:
I am getting some warnings of incomplete frames, but it’s not clear if this is due to a power issue on the powered hub or a software configuration. Despite this, the provided tools seem to work well and demonstrates some of the best depth camera capabilities I have seen (and this is my third depth camera to date).
(Nb. All resources for this post can be found on GitHub at https://github.com/mikeisted/maaxxeurope2018)
It’s been a while since my last post. My research has since moved towards the use of machine learning in UAVs and so my trusty Groundhog now sports a Jetson TX2 instead of a Raspberry Pi and an Intel Realsense depthcam for ‘deep vision’ to match it’s deep learning capabilities. But I digress… so I’ll blog more on this another time…
This post is about the DroneJam coding masterclass for autonomous UAV ‘newbies’ I ran for this year’s MAAXX Europe autonomous drone competition, held in March at the University of the West of England.
Continue reading “MAAXX Europe 2018 DroneJam Coding MasterClass”
The Groundhog is being developed to line-follow at low altitude and higher speeds. This video is of the field testing taking the speed up to 1.75 m/s. It also explains why the Groundhog now sports sunglasses.
An oval of 50mm red webbing, bends of radius approx 3m and straights of approx 15m. Testing took place in early morning with glancing sunlight on dew-soaked webbing – great for walking the dog but not so good for computer-vision.
Continue reading “Groundhog Autonomous UAV – Going faster with sunglasses”
Part of a series of videos and blogs tracking the development of The Groundhog, which was entered into the MAAXX Europe 2017 competition earlier this year.
Having successfully tested the re-written code to follow straight lines using velocity vectors for control and NED space mapping for line detection, we test it around a 50m track comprising 50mm wide red webbing – and we speed it up a bit as well.
The test turned out to be quite successful, with following speeds of 1.5m/s achieved under autonomous control provided by an on-board Raspberry Pi 3. This is significantly faster than the winning UAV in MAAXX Europe this year, which is quite pleasing!
The YouTube video shows both on-board and off-board camera footage, the former demonstrating the roaming regions of interest used by OpenCV to maintain a lock under varying lighting conditions.
Continue reading “Groundhog UAV curved line following”
Whilst I’m not into FPV, I use an Eachine LCD5800D monitor to check the view from the Raspberry Pi companion computer of The Groundhog. With the super-imposed graphics, it gives a constant view of the status of the image lock on target etc. It also has a nifty built-in recorder.
Recently the link refused to work, and after first replacing the transmitter, I realised it was actually the receiver that had failed. I decided to upgrade the receiver, hopefully fitting a new unit within the existing case.
A good briefing on the internals is to be had from the excellent YouTube video from Albert Kim at https://www.youtube.com/watch?v=P7A6fKBtbXM.
Continue reading “Replacing the receiver on an Eachine LCD5800D monitor”
Several lessons were identified here from the entry of The Groundhog hexacopter in the MAAXX Europe competition earlier this year.
Current developments are around correcting the issues so that we get a UAV successfully lapping the oval track at a minimum average speed of 1m/s.
A number of changes in approach have been made from that previously blogged. Recall the platform is based on a combination of Pixhawk/Raspberry Pi3/OpenCV/Dronekit.
- The birds eye view image transformation in OpenCV was causing segmentation faults on the RPi. Instead the position and bearing of the detected line is calculated using straight trigonometry.
- Improvements made to the ranging ROI bands to further speed-up the frame rate. This is now at a reported 50fps (which is faster than the PiCam is supplying them).
- The use of quaternions has been temporarily suspended in favour of control by velocity vectors.
As in MAAXX Europe, it makes sense to initially test on a straight line. Initial testing was conducted outdoors using red-seatbelt webbing for the line. It was not possible to fly below about 2m as the propwash blew the line away (will sort that next time!).
Initial Testing (Links to YouTube Video).
Video – Groundhog UAV initial testing on straight line.
In this last post of the series I shall overview the main program including the control algorithms for the Groundhog. Code is written in Python, using Dronekit and OpenCV all running on a Raspberry Pi 3.
As we are flying indoors without GPS and also without optical flow, we are using quaternions to control the vehicle in the GUIDED_NOGPS flight mode of ArduCopter. To be honest, I’ve not come across anyone else doing this before, so it must be a good idea…
Continue reading “Post 5. MAAXX Europe. Quaternions, control code and PIDs.”