Summary

Revolutionary photogrammetry on demand

Objective: Photogrammetry has always been considered a precise, productive and versatile method for acquisition of landscape features like topography, buildings or land cover. However, these features come at a price: photogrammetry requires heavy equipment in the field, cumbersome manual operations and is often expensive. The aim of this project is to make photogrammetry rapid, simple and inexpensive. In order to reach this ambitious goal, a lighter acquisition platform is needed and a new workflow has to be developed, while keeping the quality standards high.

Methods: Light, reliable and intelligent unmanned airborne vehicles (UAV) are now available and the focus can be set on getting the best out of hundreds of x-megapixels images. To keep cost and weight as low as possible, computations based on less accurate measurements sometimes replace direct accurate measurements. This is an exciting challenge: “How to do a lot with a limited input?”  Aerial images are produced with good regularity, even with sub-optimal conditions (little sunlight, strong wind, low temperature). Processing is streamlined in order to be able to process multiple datasets with a high throughput, a condition of success.

Results: R-Pod (rapid, rich and reliable photogrammetry on demand) is a combination of an extremely smart flying wing built by senseFly, LLC (Switzerland) and a completely rethought workflow in order to provide the traditional photogrammetric products (DTM, 3D building models, orthomosaic) with the same quality as usual, yet in a shorter time and at a much lower cost (about 50%). The time is mainly reduced thanks to the reactivity of the system: from order to flight planning to flight execution, the needed time is as little as 15 minutes. The repeatability is very good and a complete new spectrum of applications has been tested: monitoring of pedestrian streets and squares, parking lots, traffic congestions, etc. The lightness of the system allows flight virtually at any altitude, between 50 and 500 m (160 to 1600 ft) above ground, which means images with a ground resolution (GSD) between 1.5 and 15 cm (0.6 to 6 inches). Areas between 1 ha and 2.5 km2 (2.5 acres to one sq.mi) can be covered in a few minutes and are processed in a few hours.

The main features are:

  • Web-based mission planning, including cost estimation
  • Fully automatic flight execution, including take-off and landing
  • Rapid output for emergency mapping or Quality Assurance
  • Streamlined workflow with traditional photogrammetric software
  • Quality controlled photogrammetric products
  • Consistent output for virtual globes and map portals (Google Earth, Swiss Geoadmin, Open Street Map, etc.)

The specificity of the project is in the combination of new technologies and old techniques in order to obtain rich, up-to-date and inexpensive traditional products.

Conclusion: R-Pod delivers the traditional photogrammetric products in record time and at record cost. Since the acquisition platform is fully automatic, from take-off to landing included, the attention can be put on quality control during the whole process. Direct georeferencing is calculated instead of measured (as traditionally). Since the cost and the ease of operation is much more favorable, new applications are now possible, such as: survey of inaccessible areas (avalanche, flooding, landslide, bushfire), repetitive survey of a given area for monitoring at intervals as short as one minute up to hours or days and survey of less valuable areas like sand dunes or swamps.

HEIG-VD University of Applied Sciences Western Switzerland, Yverdon-les-Bains, Switzerland