My team and I at WeRobotics are partnering with the World Food Program (WFP) to develop practical coordination mechanisms for UAV deployments in collaboration. These will be developed with a range of national & local partners. In this post I want to share the basic coordination protocols we used in the aftermath of Cyclone Pam, a category 5 cyclone that devastated the islands of Vanuatu in 2015. By “we” I mean myself, the World Bank and two UAV companies from Australia (Heliwest) and New Zealand (X-Craft).
The World Bank tasked me with spearheading the UAV response to Cyclone Pam so I recruited the two companies to carry out the aerial surveys. I selected them from a dozen groups that had registered with the Humanitarian UAV Network (UAViators) Global Pilot Roster. When we landed at the international airport in Port Vila, we saw a very common scene. Military cargo aircraft filled with food, water and other relief items. Helicopters were also being chartered to support the relief efforts. And commercial aircrafts like the one that had taken us to Vanuatu were also flying in and out on a daily basis.
We clearly needed to develop coordination mechanisms that would allow us to fly our UAVs in this relatively complex airspace. So within an hour of landing in Port Vila, I organized a joint meeting with the Government of Vanuatu, Air Traffic Control (ATC), World Bank, Australian Defense Force, New Zealand Defense Force and the two UAV companies. By the end of the 1-hour meeting we had agreed on a clear set of coordination protocols that would enable us to fly our UAVs safely in non-segregated airspace. And it wasn’t rocket science.
At 22:00 every night, we would email the Australian Defense Force (ADF) our flight plans for the following day. An example of such a plan is pictured above. By 23:00, the ADF would respond with a yes/no. (They said yes to all our plans). At 23:00, we would email our approved flight plans to controllers at ATC and start programming the UAV flights. We’d get a few hours of sleep and head back when it was still dark to reach the survey sites as early as possible. This was also true for areas near the airport since we could only fly our UAVs between 6am-8am based on the agreed protocols.
Once on site, we’d set up the UAVs and go through our regular check-lists to ensure they were calibrated, tested and ready to fly. Before take off, we would call ATC (we had the mobile phone numbers of 2 ATC operators) and proceed as follows:
“Hello ATC, this is the World Bank UAV Team. We are on site in [name of location] for flight number [x] and ready for takeoff. Do we have your permission?”
After verbal confirmation, we would launch our UAVs and carry out the aerial survey. We flew below 400 feet (per UAV regulations) and never, ever strayed from our approved flight plan. The Civil Aviation Authority of Vanuatu had given us permission to fly Extended Line of Site, which meant we could fly beyond visual line of site as long as we could keep an eye on general airspace where our UAV was operating. After landing the UAV, we would call ATC back:
“Hello ATC, this is the World Bank UAV Team. We have just landed the UAV in [name of location] and have completed flight number [x]. Thanks.”
Simple and yet highly effective for the context at hand. We had the mandate, all the right contacts and we everyone followed the coordination protocols. But this is just a subset of protocols required for coordinating UAV flights. There are other components such as data-sharing workflows that need to be in place well before a disaster. What’s more, in the case of Cyclone Pam, we were working with only two professional UAV teams in a Small Island State. Just weeks after Cyclone Pam, a devastating 8.0 magnitude earthquake struck Nepal. The situation there was a lot more complex with at least 15 UAV teams self-deploying to the country.
The UN Office for the Coordination of Humanitarian Affairs (OCHA) in Nepal formally asked me to coordinate these teams, which turned out to be quite the nightmare. The Civil Aviation Authority of Nepal (CAAN) did not have the capacity or expertise to partner with us in coordinating UAV flights. Nor did UNDAC. Many of the self-deployed UAV teams had never worked in disaster response before let alone in a developing country. So they had no idea how to actually support or plug into formal relief efforts.
While most of UAV teams blamed connectivity issues (slow and intermittent email/phone access) for being unable to follow our coordination efforts online, several of them had no problem live-tweeting pictures of their UAVs. So I teamed up with LinkedIn For Good to developed a very simple Twitter-based coordination system overnight. UAV teams could now tweet their flight plans which would get automatically added to an online map and database. The UAV teams kept tweeting but not a single one bothered to tweet their plan.
To say this was problematic is an understatement. When organizations like WFP are using manned aircraft and helicopters to deliver urgent relief supplies to affected communities, they and ATC need to know which UAVs are flying where, how high and when. This is also true of Search and Rescue (SaR) teams that often fly their helicopters at low altitudes. In due course, we’ll have transponders to track UAVs in real-time. But safety is not the only consideration here. There is also a question of efficiency. It turns out that several UAV teams in Nepal carried out aerial surveys of the exact same areas, which is hardly optimal.
So I applaud the WFP for their important leadership on this matter and look forward to working with them and in-country stakeholders to develop practical coordination mechanisms. In the meantime, WeRobotics has set up Nepal Flying Labs to build local capacity around the use of UAVs and enable local responders to use UAVs safely, responsibly and effectively. All of our Flying Labs will adopt the resulting coordination mechanisms developed with WFP and stakeholders.