Hobby drones for mapping entire cities: A case study in scalability and accessibility of drone based mapping
Mapping 61 sq km with DJI Phantoms
The Revolutionary Government of Zanzibar (RGoZ) with the support of the World Bank has been developing the Open Data for Resilience Initiative (OpenDRI) with the aim of supporting evidence-based and innovative solutions to better plan, mitigate, and prepare for natural disasters. Zanzibar’s charismatic Minister for Lands, Muhammad Juma Muhammad, an architect by training, says that he’s grappling with urban growth and the needs of the tourists that prop up the local economy.
The Zanzibar mapping initiative was created to enable the governments to know the exact locations of their assets, use the data towards calculating taxes, and planning drainage. Traditionally, you would use satellite imagery to accomplish the above, but it’s not that effective in the tropics because of cloud cover and it’s not very high resolution. The other alternative is to use manned aircrafts. However, costs are prohibitive and one cannot do it every six months. Drones are an extremely nimble, low cost, and accurate way of accomplishing the same.
Low cost, scalable, and accessible mapping tools: Could DJI Phantoms be used for large areas?
With our experience at DronaMaps, we can answer that question in the affirmative. However, there are two primary challenges here, one is operational project management and the other is technical complications. In this post, we will detail the entire journey, just in case, someone somewhere in the world is considering using DJI drones towards surveying, urban planning, or other development focussed applications.
Planning
We created the plan with an assumption of using 8–9 quadcopters. The initial challenge is to decide the area and the distance the drone would go from the launch location. The other challenge is to make the flight plans detailed ensuring that no team is flying adjacent to each other. All other factors covered in our post on flight planning still hold true. The total area of interest is 61 Sq Km. The terrain here is fairly flat except at a few locations. As the elevation is not a major factor in our patch distribution, we are only dealing with the constraints borne from swathe area and line of sight restrictions.
Flight plan in patches:
Majority of the flight plans were created considering the line of sight constraints and the rest with the area distribution in mind. Thus we developed two types of execution plans. The first half was column wise and second half was grid wise.
Column-wise plan started from the northern side and extended to the southern with an average width of 180m and the length changed according to the boundary limits. On the other hand, grid wise plan was made based on the area covered in each flight keeping in mind battery and time optimisations. On- ground flight throughput for a Phantom 4 drone is about 0 .8 sq km and the maximum distance from the launch location was 500m.
Batch division for processing:
The first thing to bear in mind is that the processing boundaries and flight boundaries are different. Flight plans are created considering only the feasibility of flight, whereas, processing boundaries are decided based on the processing capability and total file size.
In our case, we created ten small plans based on the available processing infrastructure. All the column plans represented as Xc (X is plan numerical ‘c’ representing it is column plan) and Grid plans are represented as Xg (X is plan numerical and ‘g’ is representing it is grid plan).
Processing plan: The corresponding area for each plan is also mentioned:
- 1c-25c :7.80 sq km
- 26c-43c : 7.40 sq km
- 44–55c : 7.20 sq km
- 56c-60 : 4.05 sq km
- 61c-64c : 2.56 sq km
- 1g-27g: 5.6 sq km
- 28g-50g: 4.71 sq km
- 51g-66g: 5.57 sq km
- 67g-90g: 8.11 sq km
- 91g-118g: 9.71 sq km
Planning the Ground Control Points:
A total of 20 locations were identified for the GCPs. Since, the area of interest is almost flat, the criteria taken is simple, they should cover all the extremities in the plan and some should be in the middle. It is extremely critical to note that the locations should be accessible to people on the ground.
Conclusion: Statistics that matter
- Area covered: 61 sq km
- Total images processed: 112000
- Number of overlays on the map: 15
Map Layouts:
Map layouts had been prepared based on client requirements.
3D- Map with annotations:
High-resolution 3D maps are hosted online. Information of each asset is also embedded in the 3D model. The details of the plots include their unique identification number, area, the purpose the building is being used for, and the category of industry it belonged to.
