In 2018 the US Defense Advanced Research Projects Agency (DARPA) unveiled the Subterranean (SubT) challenge The aim was to develop new robotics technology that can explore underground environments – including human-made tunnel systems, urban underground, and natural cave networks – both on Earth and in space.
DARPA’s Subterranean Challenge also called on companies to find safe ways to undertake underground search and rescue in mines, caves and after natural disasters.
According to the agency, complex underground settings can present big challenges for military and civilian first responders. The locations can degrade or change rapidly over time and are often too high risk for personnel to enter. Tunnels can extend many kilometers in length and include highly constrained passages, multiple levels, and vertical shafts. While urban underground environments are often more structured and constructed out of human-made materials, they can have complex layouts that cover multiple stories and/or span multiple city blocks. Natural cave networks often have irregular geological structures, with both constrained passages and large caverns, and unpredictable topologies often stretching large distances in extent and depth.
It was crucial therefore that the robots could rapidly map, navigate, and search underground environments during time-sensitive combat operations or disaster response scenarios.
The SubT Challenge was organized into two competitions (systems and virtual), with two tracks. Teams in the systems competition had to develop and demonstrate physical systems to compete in live competitions on physical, representative subterranean courses. Eight teams have qualified for the competition final event in September 2021, where the teams’ robots have to rapidly navigate unfamiliar underground environments at a former limestone mine in Louisville, Kentucky in search of items such as backpacks and cell phones as well as mannequins (representing trapped survivors) and invisible gas.
One of these teams is the Collaborative SubTerranean Autonomous Resilient Robots (CoSTAR), formed as a collaboration between the NASA Jet Propulsion Laboratory (JPL), Massachusetts Institute of Technology (MIT), California Institute of Technology (Caltech), Korea Advanced Institute of Science and Technology (KAIST), and Sweden’s Luleå University of Technology (LTR). The CoSTAR team has finished second in the Tunnel Circuit and first in the Urban Circuit, and if it is successful in the final test, the team will be granted US$2m to go towards funding future research projects.
To succeed in the challenge, the CoSTAR team needs to be able to perform quick field maintenance for broken parts and rapidly iterate designs in between robot forays. If a part were to fail in the field, it could result in CoSTAR not being able to complete the circuit, according to the team’s engineers.
As a result, the team had to quickly design and create extra components such as brackets and enclosures. While previously they used wood, machining, or laser cutters to get the parts made, a better and quicker option was required to keep ahead of the other teams, and 3D printing was chosen for its ability to make quick iterations and fix parts at the point of need.
Initially, the CoSTAR team used a polylactic acid (PLA) filament 3D printer. However, this printer could only make parts useful for prototyping, not for eventual production. The parts weren’t strong enough to be used out in the field by a robot going into uncertain terrain with the possibility of critical components being damaged. For example, the US$7,000 light detection and ranging (lidar) sensor on top of a Boston Dynamics Spot robot (which the team named Nebula) had to be protected at all costs, so the material used to protect it had to possess improved stiffness and vibration dampening properties in order for the signal from the sensor to provide valid, usable outputs. The team’s PLA 3D printer was producing inaccurate parts with poor surface finish, so they were spending too much time sanding down the parts after they were printed.
As a result, the team used several Markforged composite 3D printing machines to manufacture a 3D printed cage for the sensor made of continuous carbon fiber reinforced material. The 3D printed cage ensured the lidar sensor could remain intact, even when Nebula took a few falls.
The team also found that 3D printed carbon fiber brackets and mounts made for the project also outperformed their aluminum counterparts and were much lighter. During the competition, the team increased the number of Markforged printers they had from one to three. Markforged then sponsored the team in 2020 and provided a fourth printer.
The CoSTAR team next plans to develop systems to enable underground exploration on the moon, Mars, and maybe even some of the moons of Saturn and Jupiter.