Field notes from the top

NOTE: This post was originally written on 7/31/14

Getting the DAO ready to fly in the house at Tres Cruces

Ground control station 3640 m AMSL

As expected, operating drones in a remote, high-elevation area brings with it a significant set of challenges. These range from aircraft performance, to battery charging, to mission planning, and more. That said, one of the major appeals of small-unmanned aircraft is their portability, making them well suited for transportation and operation in remote regions. In this post I will spend a little bit of time discussing the major challenges of operating in a remote, high-elevation locale for an extended period of time.

First, for the sake of clarification, I will describe our operating location and define what I would consider to be remote and high-elevation. We are currently operating at Tres Cruces, in Manu National Park in southern Peru. The site is 13.5 km by gravel road from the main road. The main road is also gravel and is another 30 km from the nearest town of significant size, Paucartambo (only basic hardware and electrical supplies are available in town).  We are unaware of any model aircraft/UAS parts suppliers within 3 days of travel from our location. There are no utilities whatsoever at Tres Cruces, although power and internet is available at the guard station 13.5 km away (a 2 hour, 20 minute forced march from Tres Cruces). In addition to its remote nature, Tres Cruces is situated at 3,640 m (12,000 ft) AMSL. Weather is highly variable but is typically calm and clear from sunrise until ~9:30-10:30, cloudy/foggy with as low as 50 m visibility until dusk (~18:00), then clear and windy (30-70 km/h) through the night. We are imaging steep mountain slopes at ranges up to 10 km from our flying site.

For the sake of simplicity, I will define a remote location as one with no utilities, difficult access (>2 hrs from a town of significant size), and no reasonable access to replacement parts. I will define high elevation as an area higher than 2,500 m (8,250 ft) AMSL.

I will now identify the major challenges we have identified while working in this location and then I will describe how we are working to address them.

• Decreased thrust
• Decreased lift
• Increased stall speed
• Decreased drag
• Difficult-to-predict turbulence and updrafts
• Small landing area
• Difficult battery charging due to low air temperature
• Difficult coordination of waypoint altitude with terrain elevation
• Difficulty in maintaining LOS to aircraft due to terrain
• Short flight-weather window in the early morning
• Extended travel to address basic needs

The changes in aircraft flight characteristics are pronounced and some area easier to address than others. Decreased thrust, decreased lift, and increased stall speed are unavoidable due to altitude, but fortunately most small UAS, ours included, have more thrust and lift than is necessary. This also means that stall speed is quite low at typical elevations. Stall speed has increased from ~9 to ~12 m/s, which still allows for landings at quite low speed.  Climb performance is decreased but this is easily addressed with more conservative flight profiles. Takeoff is more difficult, but launch over a receding slope mitigates this concern.

Decreased drag has shown itself to be a major concern, however. This means that the aircraft accelerates very rapidly in descents, quickly leaving the optimal flight envelope and causing the APM to have difficulty controlling the aircraft. We are still working to address this issue.

Difficult-to-predict turbulence and updrafts, especially near the ground, combined with a small landing zone, make landings quite challenging. While we have not been able to make landings simple, we have developed a few strategies to mitigate risks. First is a high landing speed. This limits the effect of turbulent air by moving the aircraft through and into clean air faster. It also provides for crisper control of the aircraft. We also use a fairly aggressive glide-slope. This limits the risk of landing short on steep hillsides and makes aborting landings easier.

Low air temperature, especially at night, complicates battery charging. Battery charging is already complicated by the need to run a generator. We typically also use the generator at night to provide light, so night is an obvious choice of time to charge batteries. Unfortunately, air temperature at 18:00 is typically only 4-5° C. This temperature is too low for LiPo batteries to function properly and prevents them from being fully charged. We have addressed this by warming the batteries in pockets and charging them in the morning when the strong morning sun can keep them warm. Fortunately due to low fuel consumption of the generator (~0.3 l/hr) we are able to do this without sacrificing nighttime power. That said, a heated room would help simplify the process. A car battery which could be charged at night during normal run-time and then be used to charge the LiPo batteries during the day would also be helpful.

Flight planning is also significantly more difficult under our current conditions. Prefetched aerial imagery in Mission Planner does not appear to include terrain elevation. This problem is compounded by our flying across a wide range of elevations while attempting to maintain a relatively constant altitude above terrain. Fortunately, our cached Google Earth imagery does have terrain data associated with it which allows us to cross-reference and determine terrain height at waypoints. This is a time-consuming process, but does allow for mission planning which maintains a safe and relatively constant altitude above terrain.

Terrain also makes it difficult for us to maintain LOS to our aircraft while it is in flight. Nearby ridges obscure LOS to the more nearby tree plots we wish to image. We have not yet identified a location that is readily accessible, which also provides LOS to the entire ridge of tree plots. As a workaround, we plan to use test flight plans to fly at what we believe is a reliably safe height above Google Earth terrain height (150 m), allow the aircraft to fly out of connection, then review flight video to adjust operating altitude based on actual observed terrain. From this we can hopefully create mapping flight plans which are low to the ground (100 m AGL) which also safely avoid terrain. We feel that routine loss of connection to the aircraft is an acceptable risk considering the observed reliability of our aircraft and the importance of the current suite of missions.

Finally, the weather provides significant challenges to our operation. We are only able to fly until 9:30-10:30 before clouds close in. In addition, regular visitors to the site to watch the sunrise typically push back our earliest takeoff time until 7:30-8:30. There is not a workaround for this other than to be ready to fly as soon as the parking area is clear and make efficient use of our time in the air. Despite the short window, we believe that we will have sufficient time to map half or more of the site in a single day. This will allow for easy and frequent re-mapping.

Finally, the remoteness of the site itself is a challenge. For example, I recently needed to seek medical attention in Paucartambo. It required an overnight stay and we lost a day of flying for a mere 1-hour visit to the clinic. Unexpected challenges such as this can easily eat away from the time available for work at a site.