UAV Challenge 2020 communications update

By | December 12, 2019
During the 2020 challenge we will need to control our ‘rescue’ quadcopter in a farm shed up to 12 km from our ground station and establish a video and audio link to the farmer in the shed, without any dependence on the cell phone network. Quite a challenge! We think that this really pushes the limits of technology that is available to the hobbyist, community groups and small businesses.
Our general philosophy for the 2020 UAV Challenge is that we want to have redundancy in our networks to avoid a single point of failure during the mission and to take advantage of the opportunities of radio frequency diversity. By having two networks operating on different frequencies we should maximise the data rates available at higher frequencies (such as 2.4GHz) while exploiting the reliable range performance and better object penetration of lower frequencies (915MHz).
We have had some significant testing success with our communications for the 2020 UAV Challenge in the last few weeks, including a strong connection at 11.2km for all three of the radio types that we have tested: our Airborne Innovations and Microhard 2.4GHz network, our trusty RFD900Xs and some promising open source ‘New Packet Radios’.
Airborne Innovations and Microhard have generously supplied us with set of 2.4GHz pMDDL2450 based radios to potentially use in the competition. We recently tested this in a point-point link test at 11.2km. The representative air platform was on top of a hill connected to a Holybro Durandal to establish a representative MAVlink connection to a ground station.   On the ground we used a video aerial systems super cannon and pepperbox antenna while the ‘air platform’ used MadMusroom antennae. We established a good telemetry and control link with very low packet loss and a fade margin of 12dB, which we are pretty happy with. These radios can be set up to mesh network, so they could be used like the RFD900x to link the rescue aircraft to the ground station via the communications relay aircraft. This set up has potential for telemetry, control, video and audio from the rescue aircraft, but we have more testing to do here.
The RFD900X radios are very capable and reliable for maintaining long range line of sight control and telemetry links with the UAV; their ability to mesh network also makes them very useful to use with a communications relay aircraft when we lose line of sight with the rescue aircraft. We used these successfully in 2016 and 2018. The big change for the 2020 challenge is the need to send video and audio from the rescue aircraft back to the ground station. We have been working on a new video encoder that can send video at very low frame rate over a low bandwidth over the RFD900X alongside the control and telemetry data. We are having some success here and it looks promising, so we have more testing to do with this setup.
The New Packet Radios are an open source design that uses the same STM32 radio module as the RFD900X but operates at around 433MHz. We think it would need a Ham (amateur radio) to operate it at the necessary power for the challenge. These radios showed good potential for a link with a good data rate between the ground station and the relay aircraft. Unfortunately these radios lack the small form factor to fit onto the rescue Quadcopter so it would only be useful for connecting the ground control station to the comms relay aircraft. For this reason, we have decided not to proceed further with testing these radios as part of our 2020 Challenge project.

Communication testing

By | October 17, 2019

Full disclosure; our current communications testing for the 2020 Medical Rescue UAV Challenge isn’t the richest material for a visual update!

We think that communications is the biggest challenge for the 2020 mission. We will need to maintain video, command and telemetry links with a QuadCopter in a farm shed 12km from our ground station. We will need to use a communications relay aircraft near the shed so that we can maintain radio ‘line of sight’ links throughout the mission. Its no small challenge to do this with lightweight accessible hardware! We are working hard on testing a number of communications options including 433MHz, 900MHz and 2.4GHz options.

We tested our current systems at 11.2km between a local hill and our flying site. With basic omni-directional antenna the RFD900Xs easily achieved and maintained a link with a 125kb air data rate. You can see below that this was a good enough to issue commands to one of our HolyBro Quads.

We have plenty of experiments left to do; we need to get some of the patch antennas set up so that we can do some experiments with our  Airborne Innovations and MicroHard radios to see what performance we get from them at similar ranges.

Finally, we have another exciting option that we are working on. Carlos has been looking into some open source 433MHz digital IP based radios. These radios are called NPR (New Packet Radio). They are a custom radio protocol, designed to transport bidirectional IP traffic over 430MHz radio links (ham radio). This protocol is optimized for “point to multipoint” topology, with the help of managed-TDMA. Bitrate is advertised as up to 500kbps (net, effective bitrate). We’ll be experimenting with these soon and we’ll let you know how we go!

Good luck to our junior team in the High School UAV Challenge this weekend

By | October 11, 2019

We are really excited that Cassandra and Elizabeth have entered a CanberraUAV Junior Team in the high school UAV Challenge and they have made it through to the competition this weekend in Queensland!

They have been working hard on developing an autonomous mission with their 3DR Solo for the main mission. They have also done some great work designing and 3D printing their own payload release mechanism and a clever protection device for their payload. Good luck to you both!

Communication, communication, communication!

By | September 27, 2019

For the last few weeks the CanberraUAV skunkworks has been busy working on solutions to the 2020 Medical Rescue UAV Challenge problems.

We see the 12 km video link between the Quadcopter and the ground control station as the biggest challenge for 2020. The Quad will be in a shed videoing ‘Outback Joe’ whilst sending video via one of our trusty Porter aircraft back to the ground station where it is being controlled from. We always try to have redundancy in our systems, so we’re working on two seperate communications links to achieve this. One of these makes innovative use of our trusty RFD900Xs and another one that uses some amazing new radios that have been kindly donated to us by Airborne Innovations and MicroHard. These radios are video and audio capable with telemetry & control and an option for manual control with an SBUS passthrough module. They claim solid video data rates at 5 miles with basic antenna; Airborne Innovations  has donated some pretty special antennae to boost that range. We’re pretty excited to get our hands on them! More on these when we have tested them.

Tridge and Matt have been working hard on developing a new video encoder that can reduce the bandwidth and lower the latency sending video back to the ground station over RFD900Xs with MAVlink. We’ve made a good start but we have quite a way to go!

We’re hoping to give the Airborne Innovations and MicroHard  pMDDL2450 radios a test this weekend, so we’ll report back then.

Here’s a taster of what else we’ve been working on for the next update!

Getting mid-air launching right

By | August 21, 2019

Peter and Jack have been working hard on repairing and modifying the Boomerang and launch mechanism to make the Quad launching reliable. They have fitted a servo hold / release mechanism for the Quad which solves the problems that we had last week where the Quad was blown off the Boomerang by the prop wash and air flow. Combined with Peter’s ‘punch’ launch code this is now working well and reliably. One step closer to saving Outback Joe in the 2020 Medical Rescue UAV Challenge!

Boomerang 60 mid-air launch development and Hex HereFlow optical flow testing

By | August 15, 2019

This week we moved from the Hobbyking Bixler 2 as a launch platform for our HolyBro QAV250 Quadcopter onto a bigger platform, a Boomerang 60. The Boomerang is a conventional ‘pull’ configuration training aircraft, compared to the ‘pusher’ electric-glider configuration of the Bixler. We made this transition because we think the Boomerang is more representative of the type of airframe we will use for the support aircraft in the 2020 Medical Rescue UAV Challenge. We learned a lot in this short and unsuccessful mission! We had underestimated the effect that  the propeller in front of the QuadCopter would have, as you can see clearly from the video. Back to the drawing board for the mounting system!

We also tested the new Hex HereFlow combined optical flow and LIDAR unit over CAN on another of our HolyBro QAV250 QuadCopters. In the 2020 Medical Rescue UAV Challenge we intend to drop the QuadCopter near Joe’s shed then navigate it into the shed to achieve video and audio communications with Joe. To do that we need to be able to navigate the QuadCopter precisely from outside the shed into and then inside the shed. We found that the LIDAR struggled to maintain accurate height in the bright Australian sunshine, which  is likely to be a problem in the challenge. We found that the combined sensors of the HereFlow worked quite well in the shade outside. As you can see from the video, the QuadCopter maintained a decent 1 metre circle in reasonable wind without GPS data, but maintenance of the set altitude was still a little erratic. The performance in the shed was better again, as you can see on the video. Our test validated the manufacturer’s known limitation that the LIDAR doesn’t work over two metres altitude. Our tests indicated that it has limitations in high illumination levels too, but we didn’t do enough testing or have a light meter to properly quantify this. Overall the HereFlow seems like a promising tool but we need to augment it with a better LIDAR rangefinder. We’re planning to use a Lighware LW20, so watch this space!

More mid-air launch testing

By | August 7, 2019

Another exciting week for Canberra UAV and our preparations for the 2020 UAV Challenge. We conducted three more tests of a QuadCopter launching off the back of a fixed wing aircraft, the Hobbyking Bixler 2. We are running this particular Bixler 2 on 4S batteries rather than the normal 3S because it needs the extra ‘oomph to carry a QuadCopter that weighs as much as the Bixler itself!  Last week we had noticed that the ‘copter spin up and launch was quite slow compared to the requirement for mid-air launching. Peter has been working on a ‘Punch’ launch mode for auto take-off to solve this problem; essentially a short burst of full throttle with stabilisation to seperate the ‘copter quickly and cleanly from the support aircraft. This ‘punch’ mode seems to work well, but on our limited number of tests we have seen that it is still difficult to clear the Bixler’s pusher prop cleanly. The video graphically shows what can happen when you don’t!

Test 2 is a particular highlight; it seems like Peter has also coded ‘victory roll after take off’ into the ‘punch’ code!

Next week we are moving up to another work horse of the Canberra UAV fleet as the support aircraft; the Boomerang 60. This will also involve a new version of Peter’s successful 3D printed launch rail and some refinement of the ‘punch’ take-off.

We have now proved that air-launch is technically possible and our hardy HolyBro PixHawk 4 QAV 250 QuadCopter has survived every attempt, but we need to make it worthy of a UAV Challenge attempt.  During the Challenge the Quad will also be carrying a camera, a microphone and some significant communications gear, so the Quad’s thrust to weight ratio won’t be as healthy as in these tests; this makes it even more important to make the launch stable, predictable and reliable.