FAA awards first approval for drone swarm testing

Lawrence Livermore National Laboratory’s (LLNL) Autonomous Sensors team has received the Federal Aviation Administration’s (FAA’s) first and — to date — only certificate of authorization (COA) allowing autonomous drone swarming exercises on the Lab main campus.

These flights will test swarm controls and sensor payloads used in a variety of national security applications. 

Autonomous drone swarms (sUASs) differ from those used for entertainment purposes that you might see at a baseball stadium because autonomous drones are designed to operate spontaneously and independently in real time.

Drone shows typically use a 3D animation tool, Blender, to create shapes and recognizable characters generated in advance from a pre-set series of commands generated and controlled by a single operator.

“The Lab has been exploring how to apply cutting-edge artificial intelligence (AI) and machine learning to its autonomous sensors, but we couldn’t actually field-test those tools,” said Brian Wihl, systems engineer at the Lab and project lead for this initiative.

“Receiving this approval enables us to take the next step in our research. We’ll be able to apply swarming technology across several national security mission spaces to see how the swarms learn and respond in real-time.”

Past field-testing authorization required one operator per drone; this COA will allow one operator to manage multiple drones at one time.

Autonomous drones developed at the Laboratory utilize a portfolio of Laboratory-developed software and machine-learning tools designed to teach the drones how to communicate and work together in real time, respond to environmental degradations and course-correct when issues arise.

The FAA COA will allow for the Autonomous Sensors team to operate up to 100 drones during the daytime at any time over the next two years.

“This broad-use-case approval will allow us to flexibly and rapidly develop our tools and expand capabilities,” said Jacob Trueblood, Autonomous Sensors electrical hardware team lead and principal investigator for this initiative. “We’ll be able to benchmark the software currently available and recommend future updates.”

Previously any single-pilot swarm test had to be able to fit inside the on-site drone pen, which limited the scale of the test. “Now we can physically test new, complex swarm control algorithms rather than be limited to simulations,” said James Reimer, Autonomous Sensors robotics team lead.

“This approval applies to nearly all work in the Autonomous Sensors Program area, including radar work and undocumented orphan wells as well as multiple research areas across the Lab, including nuclear non-proliferation, chemical and biomarker detection, and energy and homeland security,” Wihl said.

An immediate focus for the swarm work at the Lab is using multiple ground-penetrating-radar-equipped drones for explosive hazard detection.

Trueblood is currently leading the undocumented orphaned wells project to locate undocumented, abandoned gas and oil wells in Osage, Oklahoma.

“Being able to send a swarm to rapidly search an area to detect leaking methane instead of just one drone at a time will be a significant workforce multiplier,” Trueblood said.

To prevent drones from leaving the Laboratory’s campus, the autonomous drones have a soft geofence, a hard geofence, and an emergency stop. A geofence is a predefined virtual boundary to constrain the flight area of a drone.

Each drone uses its GPS to monitor if it is approaching or has breached a fence. If a drone reaches the soft fence, it will immediately return to the home base from where it took off; if a drone reaches the hard fence, its motors will automatically shut off. An operator also can trigger an emergency stop to immediately shutdown the drone’s motor at any time.

“An operator can choose between multiple emergency commands, including hold, land, return to home and disarm to shut down the drone's motors. The emergency commands can be sent to individual drones or the entire swarm at any time,” Reimer said.

To keep onsite Laboratory employees safe, the Autonomous Sensors team will place barricades and signs at the entry points and/or install access-controlled gates at the field-test entry points. When operating the sUASs, the team also will notify the LLNL protective forces division to limit access to the field-test areas and use the visual observers and ground crew.

The sUASs will operate at a ceiling of no more than 400 feet, which should not present any hazards to aircraft in the area. “We really don’t have any issues with other aircraft flying low at our campus unless they have coordinated with the Lab’s safety officer, but we’ll still monitor for traffic,” says Kathy Brown, LLNL’s aviation safety officer.

“We’ll also ensure the people that are involved in field testing the swarms are properly trained in sUASs safety. We won’t operate the sUASs above anyone, including those involved with the operation, and we’ll ensure that everyone remains outside of the soft and hard geo fences,” said Doug Eddy, Livermore Field Office senior technical safety adviser who completed and submitted the COA application and interfaced with the Department of Energy headquarters and the FAA to guide approval of LLNL’s application.

“The work the Livermore team put into obtaining this approval required significant collaboration across multiple teams and will make a real impact in taking this work to the next level. The approval the Autonomous Sensors team obtained for swarm behavior field testing will also help other groups at the Lab exploring questions around swarm behavior, artificial intelligence and machine learning,” said Reg Beer, program leader for Defense Systems at LLNL.

“We expect to obtain multiple COAs in the future as we push the technology forward and take it where it needs to go.”

This initiative received funding from multiple projects, including Skyglass under the Office of Naval Research (ONR) and from the Lab’s Autonomy Support Group.