Combining drone telemetry data with 3-D laser mapping data during the investigation of the Amtrak Cascades derailment reduced police on-scene processing time by up to 80 percent (compared to traditional methods).

When an Amtrak train derailed onto a major highway near DuPont, WA use of drones helped the State Police quickly and accurately gather the evidence needed to clear the roadway and resume normal traffic.

Date Posted
08/15/2019
Identifier
2019-B01389
TwitterLinkedInFacebook

Drone Use During the Amtrak Cascades Derailment and Response

Summary Information

On the morning of December 18, 2017, Amtrak train 501 derailed near DuPont, Washington, causing railcars and one locomotive to fall onto Interstate 5 (I-5), hitting several passenger cars and shutting down the freeway. I-5 southbound was ultimately blocked for 57 hours, an impressive accomplishment considering the necessity of surveying the incident for the investigation by the National Transportation Safety Board (NTSB), the process of removing a 270,000 pound locomotive, and the need for inspections and repairs before the area was cleared to open. Approximately 65,000 vehicles per day had to be diverted on this southbound route and forced onto local roads and highways. While efficient clearance time was due in large part to the pre-planning and communications efforts by the I-5/Joint Base Lewis-McChord (JBLM) Corridor Joint Operations Working Group, the use of drone technology to collect crash data significantly reduced the time needed to open up the road.

Methodology
Following the Amtrak derailment and closure of I-5, the Washington State Police (WSP) utilized unmanned aerial vehicles (UAVs) to collect the information required for the NTSB investigation. Two detectives deployed 1 UAV on a total of 25 missions, collecting 1.1 billion points of data within about 5 hours. Additionally, 3D scanners were also deployed: southbound on I-5 from Highway 116 towards the trestle and then another group ran northbound. The 3D scanners allowed WSP to go under the trestle, between the cars and down the hillside.

Findings

  • It was demonstrated that the average on-scene processing time with the UAV was 36 minutes as opposed to an average of nearly 180 minutes with traditional tools. What would normally take a couple of days to investigate, instead took five hours, allowing for the locomotive and debris to be moved quickly and the road to be opened within 57 hours.
  • The 3D scanner and UAV provided the ability to gather not just the evidence assumed to be important but also mapped the entirety of the crash and provided information that may be deemed useful at a later time.
Goal Areas
Results Type
Deployment Locations