Severe Minibus Accident Documented for Crush Analysis

After a serious minibus accident occurred in Maryland this June, the Anne Arundel County Police used Pix4Dmapper and a DJI Phantom drone to document the scene for investigation…

  • Organization Anne Arundel County Police
  • Country USA
  • Industry Forensics, Law Enforcement
  • Project Date June 2016
  • Project Manager Greg Russell
  • Project Size 2.26 Acres
  • Equipment DJI Phantom 3 Professional
  • Number of Images 154
  • GSD 1.6 cm

Project Description

GCPs and the merging method were used in order to solve the issues of missing image geotags and the multiple flights. When separating into subprojects, care was taken that each subdivision contained at least 4 GCPs and shared a portion of images with the adjacent subdivision. When merging the subprojects, manual tie points were added in Pix4Dmapper rayCloud editor in order to better coincide the subdivisions.

Due to the presence of several complex objects at low altitude such as trees, poles, tanks, container, buildings, trucks and several “difficult” surfaces to match such as water and photo-voltaic panels, the rayCloud editor and mosaic editor were used to do final polishing of the DSM and the orthomosaic.

Data Acquisition

Images were acquired by an AirVision NT4-Contras. The on-board GPS had some problems and so the autonomous flight plan was not used. Therefore, both the UAV flight and the image acquisition were done manually.

This resulted in the missing of image geolocation and inhomogeneous image coverage which may lead to insufficient overlap and difficulties in feature points matching. Also, without the automatic execution, there were three different flights (instead of one) that need to be merged in Pix4Dmapper.

For acquiring accurate global geo-position without the geotags of images, 8 ground control points were placed and measured with RTK in the survey area.

Achieved Result

Despite the above mentioned acquisition issues, the thorough use of Pix4Dmapper allowed the achievement of high-resolution (less than 3 cm GSD) and extremely accurate results over the studied area. The derived DSM was successfully used to calculate the volume of the sand/gravel stockpiles with an accuracy requested by the customer, who was very happy with the work and wanted further surveys in order to constantly monitor the evolution of the stockpiles through time. For this reason, we decided to take full advantage of the high resolution and accuracy of georeferencing, and are going to plan multi-temporal surveys with associated differential calculations of the stockpiles’ volume through time with the use of Pix4Dmapper.

Advantage of the New Technology

Using drones to document this accident scene was very advantageous, in both time spent and cost. Speed is crucial for forensic cases, in order to capture as much accident information as possible before cleanup starts, or before evidence is damaged, in adverse weather for example. Speed also influences the amount of time a road or highway needs to be shut down. For accident cases similar to this one, the highway could have been shut down 3 hours (e.g. using LiDAR to document the scene) instead of 20 minutes (using a drone), causing more damage to circulation and even to the economy.

Another advantage of drone documentation is the fact that taking images or video clips with a drone records detailed information of an area of interest without intruding on police site investigation. Within a few hours, these drone images can be turned into a fully reconstructed scene by with Pix4Dmapper Pro software. Exported 2D orthomosaics and 3D models provide proof and evidence for the collision investigation report presented in court.


Author: Gaetano Di Achille