While most of the world is recuperating from the excitement of the Tokyo Olympics and Paralympics; others have been gearing up for the ‘Robotic Olympics’.
The World’s Most Challenging Robot Competition
The Subterranean (SubT) Challenge, held by the United States Government Defence Advanced Research Project Agency (DARPA), challenges teams to develop autonomous robot fleets that can navigate and map complex and demanding underground environments.
While navigating and mapping, the fleet also needs to identify and geo-reference the location of artifacts like backpacks, cell phones, trapped survivors, and even invisible gas. The competition aims to encourage the development of technologies for improving the situational awareness of dangerous areas before humans are required to enter.
DARPA’s current SubT Challenge began in 2018 and has conducted three preliminary circuit events in tunnel, urban, and cave subdomains. The teams are competing for a share in $5 million in prize money and the prestige of winning the most challenging robotic competition.
The final event is currently underway, running September 21 to 24 in the Louisville Mega Cavern in Kentucky. The course is a combination of all three environments and is unknown to the contestants. Only one person from each team is allowed to operate the entire fleet, which will be exploring well beyond line of sight and standard communication range to cover the course. Teams have only 60 minutes to deploy their robots and report the location of found artifacts to DARPA. The team will receive a point for each artifact disclosed with an accurate geo-referenced location (within 16 feet 4.9 inches or 5 meters).
According to one contestant, the course is generally very complex and includes narrow corridors, rooms, long tunnels, stairs, multiple floors, shafts and very large. It can afford to be large, as DARPA constructed the course within the Mega Cavern, which, as the name suggests, is considerable – over 4 million square feet (371612.16 m2), and formed from a massive limestone quarry.1
Fortunately for one of the robotic contestants, underground mines are where it thrives! Initially developed for use in hard rock mining, Hovermap can easily handle the narrow confines and unique obstacles typical underground.
Hovermap is one of the robots that make up the CSIRO Data61 team, with staff and technology from Emesent (the producers of Hovermap), CSIRO, and Georgia Tech.
The Australian Team
With groups from two entities based in Brisbane, Australia (CSIRO and Emesent), the CSIRO Data61 team is Australia’s representative in the competition finals. They’ve already traveled to the USA to compete and were selected and funded by DARPA.
In the final circuit, the CSIRO Data61 team will run an advanced fleet that includes two Hovermap-enabled drones, as well as two Boston Dynamic Spots (walking robots) and two BIA5 ATR unmanned ground vehicles (UGVs).
The fleet is equipped with advanced autonomous capabilities, including autonomous exploration. They also collaborate among themselves to share tasks to maximize the coverage of the environments.
Capabilities of the Fleet
As they will be working beyond line of sight and standard communication range, the fleet requires advanced autonomy capabilities. Below are some of their many skills.
Autonomous exploration, where the drones define their own missions by navigating into unexplored frontiers with the objective of exploring and searching the environment.
Single mission execution for the drone, initiated by the launch command, which will autonomously start Hovermap, do all the pre-flight checks, arm the drone, release the latches, take off, explore, detect, report, and return to home, and land.
Autonomous exploration allows the drones to define their own missions and navigate into unexplored frontiers to explore and search the environment.
Single mission execution for the drone, initiated by the launch command, which autonomously starts Hovermap, does all the pre-flight checks, arm the drone, release the latches, take off, explore, detect, report, and return to home and land.
Building a mesh network based on the communication nodes deployed by the BIA5 ATR UGVs.
Auto syncing to the mesh network to send the data to the Ground Control System (GCS) and DARPA.
Share data with the other fleet robots in near real-time, providing each robot and the Ground Control System (GCS) a good global map.
Automatically merge data shared between robots.
Search for, clarify, and geo-reference artifacts using additional perception systems and onboard AI.
Automatically geo-reference from some ground control points laid by DARPA.
Advanced return-to-home failsafes.
Most of these technologies Emesent will be commercializing shortly, according to CTO and Co Founder of Emesent and Emesent’s lead in the team, Dr Farid Kendoul.
“Having a fleet of driving, walking and flying robots that are achieving complex missions autonomously and collaboratively is a major technological milestone and an important step towards using autonomous systems for saving lives and helping people.”
“This is the result of hard work for many years by many Emesent team members. A big thank you to all the other people who helped with this project.”