The Defense Advanced Research Projects Agency (DARPA) of the United States is responsible for conducting research and developing technologies for military use. Universities and private research companies throughout the world have all collaborated with DARPA to bring forth technological advancements. Many of these projects go on to be used for military purposes, while others are used in commercial endeavors. When DARPA is looking to develop a concept, one of the ways they attract collaborators is by hosting challenges.

One such challenge that began in September of 2018 and finished on September 24, 2021, was the SubT Challenge. As explained on the challenge website, “The DARPA Subterranean or “SubT” Challenge seeks novel approaches to rapidly map, navigate, and search underground environments during time-sensitive combat operations or disaster response scenarios.” DARPA presented the challenge with substantial incentives. Prizes from 1st-3rd place would be $2M, $1M, and $500K. Additionally, a virtual competition would award 1st-3rd place prizes of $750K, $500K, and $250K.

Winning the $2M 1st place prize was a team of students from the Norwegian University of Science and Technology led by a Ph.D. candidate from the Department of Engineering Cybernetics, Paolo De Petris. The team’s design, called CERBERUS, is a system of coordinating drones that can explore and map an urban subterranean landscape. For Paolo and his team, there were several obstacles to overcome. The primary one is that subterranean environments are often denied GPS information and wireless connectivity, both critical for autonomous drone operation.

To overcome this obstacle, Paolo developed a ground rover that accompanies the drones to provide continuous connectivity and positioning. The ground rover has what Paolo describes as “a high-gain antenna and an optical fiber connection to the base station”. As the drones move through the landscape they let out a series of “breadcrumbing” wireless signals that communicate with the ground rover and base station. The next part of the challenge was to develop drones that can traverse underground environments.

For this Paolo came up with 2 primary models, an aerial and legged drone. The aerial drone can be used to scope ahead and fit into tight or inaccessible spaces. The rugged-legged drones can crawl through tunnels to generate maps, and have the benefit of extended battery life. Both drones can be outfitted with an array of sensors to complete mission objectives. The drones are also programmed with anti-collision software, a must for an autonomous approach. “To enable autonomous exploration of subterranean environments, team CERBERUS focuses on multi-modal and multi-robot perception, as well as path planning for exploration and area search,” states Paolo’s paper on the project. “A unified complimentary onboard perception solution is developed that fuses LiDAR, cameras, and inertial sensor cues. This enables resilient localization and situational awareness even in conditions of perceptual degradation such as geometric self-similarity, darkness and the presence of obscurants…In brief, each robot can detect objects of interest in the field, and with map optimization, refine their position estimates back at the Base Station.”

DARPA has made no announcement as to how they plan to utilize Paolo’s CERBERUS drone platform. But, with the winning entry, Paolo has the opportunity to continue his research project. He has already begun a new project called the RMF-Owl. Though not part of a DARPA challenge, the RMF-Owl is a continuation of CERBERUS. It is a collision resistant drone for autonomous subterranean exploration.