The key objective of the project was to trial the ability of aerial drones to provide “eyes in the sky” to assist navigation of the supply ship through the thick sea ice, and as a secondary objective to provide an accurate aerial mapping of Casey Station.
To assist with navigation our requirements were to fly above the ship while providing a live vision feed to the bridge to enable the ship’s Master to see ‘leads’ in the sea ice. These cracks and weak points help expedite the ship’s progress and reduce the risk of it becoming trapped.
Drones have the potential to be quick to deploy (less than 10 minutes from initial instruction to delivery of video to the bridge), reduce costs and most importantly avoid risking lives in adverse conditions. Prior to departure we learned that the British Antarctic Survey were undertaking exactly the same trial on their icebreaker.
A range of key challenges were identified prior to the departure, key amongst these were the differences in how GPS and compasses work at such southerly lattitudes (over 100 degrees of compass declination at Casey Station), which had caused the drones of previous attempts by others to fly away, not an ideal situation on a moving ship.
Given the isolated location, extreme weather and other challenges, it was important that sufficient drones and backup equipment were brought along in order to guarantee a reliable and independent operation. To this end five different drones were taken, including a number of attrition aircraft which thankfully were not needed. Back-up UAVs were provided by Ground Effect Aviation, the company where James had undertaken his aviation training many years before. “We were very excited and proud to see James operating in such a challenging UAV environment”, says Matt Rayson, Chief Remote Pilot of Ground Effect Aviation. “James was our first graduate when we began the UAV training program, and his continued success provides a great example of what the right training and business planning can do”.
Tests Prior to Departure to Hobart
Our multi-rotor drones were tested in a range of situations prior to the departure from Hobart to ensure their best chance of success under these new and uncertain conditions.
In order to calibrate their onboard sensors, many drones rely on being completely stationary when first starting up. Obviously on a moving ship this is not an option. As an initial quick test the ability to operate from a moving vehicle was first trialed driving in a car as an easy confirmation that this was possible with the platforms we were looking to deploy.
The next obstacle was magnetic field interference, which initially prevented starting up in close proximity to metal structures, and would obviously be an issue when launching from the deck of a steel ship. This was identified early on as a problem to be resolved, and we reached out to the manufacturers in the hope of being provided a custom firmware to skip this calibration step upon startup, but they were unable to assist and so we went ahead trialling our own workarounds.
The impact of the cold on the battery endurance is well documented and it was assumed that there could be up to a 50% reduction in the battery life during the flight. The UAV’s battery bay was insulated to reduce this impact prior to flight in cold environments. This was later verified to have been successful, in fact we needed to reduce the insulation as they were getting a little hot.
The ship’s high-power communications equipment could also have posed a problem with regard to radio frequency interference. In the usual case on dry land the default behaviour of a drone when it loses its radio link is to execute a ‘Return To Launch’ (RTL). In the shipborne case this would have resulted in the loss of the aircraft as the launch point is, by then, some distance behind in the ship’s wake. To ameliorate this risk we amended our flight procedures to regularly update the RTL ‘home’ point to match the ship’s current position as it moved forward so that a loss of signal would still keep it close rather than flying off into the distance, and thereby give us a better chance of recovering the signal as it descended to land nearby.
The day before departure a trial flight was undertaken on the ship at dock in Hobart, Tasmania. As expected the ship’s steel construction interfered with the compass and the drone’s safety arming checks prevented the motors starting up when placed on the deck. Further tests were undertaken in different areas of the ship to determine where this calibration issue would occur, and it appeared to encompass the entire ship, which was not unexpected.
At this point it was clear we would need to disable the compass sensor for initial arming, which was not as simple as flying in ‘non-GPS’ mode, as the drone still required a successful compass reading in this mode, even without GPS. Through further testing we discovered a workaround whereby we could manually disable the sensors, allowing us to successfully arm and takeoff in full manual mode.
Testing Parameters On Board the Moving Ship
Informed by our Hobart dock tests and the prior knowledge of other UAV operators losing control in GPS mode, a succession of manually controlled test flights were scheduled to introduce increasingly complex flights in stages as each risk was tested and minimized.
Initial flights were conservative and below the level of the catch nets that surround the helideck. This first hover test was to establish that the drone would respond to the controls well and not behave erratically with the compass variation and radio frequency interference. It was also used to test battery endurance and temperature compared to a similar flight in mainland Australia. The insulation of the battery and pre-heating the batteries prior to use was very successful and there was no apparent reduction in flight endurance.
The second flight climbed to 30m above the helideck to check control responses and battery endurance. The ship was stationary at this stage in case the return to home function was initiated by radio interference. All flights were undertaken without GPS assistance. The potential for erratic flight with the variance between compass and GPS was considered too great to test off the ship.
Successful Results in Real World Conditions
The previous flight testing allowed subsequent flights to commence in full operational conditions of up to 75m above the ship in a 17kt wind and falling snow. Results were good with vision of the sea ice clear in the high definition video to an estimated 20km. The video feed was used to inform the passage of the ship, thus proving itself successful in a real world scenario.
The trials demonstrated that the ability to launch from a moving ship with a multi-rotor UAV in high wind and snow was possible and provided useful intelligence to the ship’s crew for ice navigation on the journey to Casey Station.
Further flights were undertaken at Casey Station to test the compass calibration and flying characteristics with GPS mode enabled at extreme southerly latitudes. Although the UAV’s flight manual expressly states that flying in GPS mode is the polar regions is not advised, a successful GPS hold flight was completed in very controlled conditions, ready to fall back to manual control if necessary. We were very happy with the performance of our multi-rotor drones in a range of very challenging environments and were given some fantastic positive feedback from the Australian Antarctic Division staff and crew of the Aurora Australis.
Australian Antarctic Division Project Manager Dr Sandra Potter notes ‘We knew there’d be a raft of technical challenges. Australian UAV systematically decoupled and worked through each of these, establishing that UAVs can (i) be safely deployed from our icebreaker and (ii) used to capture sea ice imagery of ship navigation value. Indeed the imagery and the ease with which it was captured exceeded our expectations. Furthermore the trial has highlighted the potential benefits of UAVs in supporting other Australian Antarctic program activities – uses long known to us but little explored to date.
Mapping Casey and Wilkes Stations
Our secondary objective in Antarctica was to trial the use of fixed-wing UAVs in the aerial mapping of Casey Station and surrounding terrain. Australian UAV’s core business is in providing accurate aerial mapping, and we were excited to see if we could undertake it in Antarctica with all its challenges. As with mainland Australian clients the delivery of a high resolution aerial image map is invaluable for site planning and documentation, and there is a lot of potential for further projects including mapping of fauna, ice change and wider landforms.
The airspace around Casey Station operates under the same regulations as the rest of Australia, such as the maximum 400ft altitude above ground for UAV operations. In order to minimize potential disturbance impacts on wildlife, inquiries were made to the Civil Aviation Safety Authority (CASA) prior to departure to request an exemption based on the extremely low (i.e zero!) risk that there would be other aircraft suddenly present in the area. However advice provided by CASA indicated that we would still require an Area Approval and NOTAM to fly above 400ft there.
Based on our experiences with the multi-rotor sensors, it was with great trepidation that one of our senseFly eBee aircraft was launched at Casey Station — with the manual controller within quick reach! As well as the potential for compass and GPS issues the battery performance impact in the sub-zero conditions was more likely to affect the eBee given its smaller battery, lower battery load (i.e. less self-heating), and reduced ability to insulate it from the cold given the small battery enclosure. Flights were planned with conservative endurance until the influence of the cold on battery life could be established.
The first flight was planned far enough away from the Southern Ocean, wildlife and other hazards to ensure that there was ample area to execute an emergency spiral landing should the compass magnetic offset cause a control problem. Monitoring its performance tracking toward the first waypoint it, was clear that the eBee was flying well. Under favourable weather conditions we completed our mapping of the Station.
At least on this occasion, despite the risks, the methods for flying the eBee in Antarctica proved to be no more complex than on the Australian mainland, which is a testament to this very capable platform.
Results of Aerial Mapping
The third major potential hurdle after the compass and cold related to the ability of the image processing software Pix4D to deal with the large areas of quite featureless snow. To help this our flight planning had been conservative with 75/80% overlap, and the software managed to find sufficient features even in the expanses of uniform snow to resolve a complete high resolution map and 3D terrain dataset around the Station.
Our team demonstrated with great success that multi-rotor drones can be of assistance to ships navigating sea ice in very challenging conditions, and also that fixed-wing aerial mapping has many potential positive applications in Antarctica (subject to environmental permitting). We would like to thank the Australian Antarctic Division for hosting us on what was an incredible journey, and in particular Sandra Potter for having the foresight to push this innovative use of drone technology within the organisation.
About the Author
James Rennie is the Chief UAV Operator at Australian UAV based in Melbourne. Following a 17 year career in environmental science James saw the value of economical aerial data capture to his existing clients across a range of industry sectors. He founded Australian UAV in 2013 with a focus on safety and professional data deliverables. The business has since grown to cover the eastern States with offices in Victoria, NSW, Queensland and Tasmania, with more locations on the way soon.