UAV’s to the Rescue—Defining a Purpose-built Drone for Collapsed Building Search & Rescue

The Continental United States has not experienced a really damaging earthquake since the 1994 Northridge, California quake. Yet around the world, large quakes have been devastating, particularly in Japan. Many experts believe that the US is well overdue for a bad earthquake. California is generally regarded as the most vulnerable area in the lower US, on the Western edge of the Ring of Fire it shares with Japan, and well populated with active faults, of which the San Andreas is probably the best known. The state has worked diligently to shore up infrastructure, but despite these efforts some structural collapse of older construction is inevitable, and experts do not rule out major events in other areas of the country.

Collapsed building search and rescue is extremely difficult and time-consuming. A partially collapsed building can be simply too dangerous to enter directly. In these cases, shoring is required to basically construct a tunnel, or a building can be dis-assembled from the top. The assignment requires great technical skill, and while many metropolitan fire and rescue organizations have developed organic S&R special units, the job often falls to one of the 28 Urban Search & Rescue Task Forces distributed throughout the United States.

In the event of a major catastrophe, there can be many collapsed buildings, and few resources equipped to mount the necessary technical rescue. It can be necessary to concentrate the available resources on those structures where there is clear evidence of live victims present. Other buildings may have to wait, motivating civilian volunteers make efforts on their own, which often increases the casualty count.

Agile, multi-rotor Unmanned Aerial Vehicles (UAV’s) can provide an answer. These can enter through small spaces and search structures too dangerous for humans to enter. They can pinpoint the location of victims within the structure, and potentially make a rough assessment of a victim’s condition. A small army of such drones can search a large number of buildings, targeting those that need special attention from precious S&R resources.

As with everything in search and rescue, there are challenges. Command communications and First Person Video (FPV) for most UAV’s are necessarily at higher frequencies which can support higher bitrates. However, higher frequencies do not penetrate solids well, such as the walls of a collapse. A UAV would likely lose the video that the pilot needs to navigate, and the command signals from the controller that tell the UAV what to do. Hazards can exist such as exposed wires which may be difficult to see on low resolution video. The pilot’s view is typically limited to what is in front of the UAV’s FPV camera, he/she cannot see hazards on the periphery, or potentially victims.

What is needed is a UAV outfitted with highly specialized equipment specific to the collapsed building search mission. The following list is based on research and the Author’s own experience and training as a volunteer with the Community Emergency Response Team (CERT). The first few items are “must haves” for the mission, the rest are for consideration. Clearly most of these features add weight and require power from the limited supply available, both of these can affect the endurance (flight time) of the UAV:

  1. Smaller and ruggedized. The UAV should be able to survive a crash and become airborne again without having any of its systems compromised. This implies that sensors and other equipment should be rugged also. Part of the UAV’s value is the ability to fit through smaller openings, which will be impaired by extensive protruding sensor equipment.
  2. Autonomous. As described above, the UAV must not rely on control from an operator on the exterior. In his book Army of None, Paul Scharre declared that “fully autonomous systems sense, decide and act entirely without human intervention. Once the human activates the machine, it conducts the task without communication back to the human user”. Autonomous UAV’s are viewed with a great deal of apprehension, especially in the development of warfare-related systems. For the S&R UAV, autonomy is essential. DARPA’s Fast Lightweight Autonomy program is currently developing UAV systems for this application. A San Diego based startup Shield AI is building an Artificial Intelligence-based platform for supporting first responders with autonomous building search capability. These systems are the core component of the S&R UAV, without autonomy the system cannot fulfill the mission.
  3. Collision avoidance. This has to be mentioned, although in this context autonomous operation can be assumed to include it.
  4. Victim Detection. Video is not necessarily the most reliable method for identifying humans using facial recognition and other techniques. The actual status of the victim may not be apparent in a video. In this case infrared detection may be preferable, since video is not required for navigation (the Shield AI system relies on LIDAR), and rescuers need to know little more than the temperature and overall size of a mass to decide that they should affect an entry and initiate formal search/technical rescue.
  5. Audio detection. Sound can carry a long way. Rescuers working a building collapse by peeling the debris back often stop all operations and go quiet so they can listen for sounds made by live victims. The UAV may be able to filter its own engine/propeller noise, or perhaps land and go quiet periodically to record sound. It may also make sense for the UAV to play a recorded announcement, inviting any nearby victims to speak.
  6. Breadcrumbs. The UAV might be programmed to return to base as soon as one potential live victim is located. If the UAV can mark it’s passage at periodic intervals, either with fluorescent paint or dropped objects, the rescuers will not have to wait for an extensive post-analysis/mapping exercise to begin rescue operations.
  7. Marker beacons. Where the UAV encounters possible victims, it might drop a low-power flashing LED beacon. Even a low-power light can often be seen at a great distance in absolute darkness. Like breadcrumbs, it can aid the rescuers in locating victims without waiting for post-analysis.
  8. Low Frequency Communication. This may not be possible due to size requirements for propagating antennas. At very low frequencies data rates are also low. The UAV may be able to communicate an event, or in some cases a low-resolution JPEG or infra-red image, or possibly an audio track.

Clearly the UAV described here is purpose-built for the task, and incorporates guidance software that also controls when certain event stake place, such as stopping/listening. These UAV’s have to be considered expendable, and they will no longer be in the “inexpensive” class. However, their use in catastrophe situations could save many lives. In any case autonomous operation such as that being developed by Shield AI is the lynchpin of this project. Without it the mission cannot be performed.

This type of development project is the future of UAVs. UAS systems need to be purpose-built from the ground for specialized use cases, and not adapted from commercially available systems. When UAVs with a purpose become the norm, they will get the larger acceptance that has been lacking, and become integrated into the framework of civilization.