Source: Wettergreen, David; Cabrol Nathalie; Baskaran, Vijavakumar; et. al. (2005) Second Experiments in the Robotic Investigation of Life in the Atacama Desert of Chile (Link)
Autonomous Surface Navigation (III-B-3)
The ability to demonstrate that an autonomous vehicle, or systems of vehicles, can remotely traverse over long distances enables prepositioning of assets prior to human use, as well as broader and more robust precursor missions. Artificial intelligence is sure to be a key factor in this development, but additional questions, such as: “What infrastructure can be put in place to provide the most efficient and least error-prone autonomous surface navigation?”, “What is the maximum distance, over what terrain, NASA might reasonably expect these systems to need to traverse?”, and “Does this approach require a collective effort, or can many individual missions converge into a general autonomous surface navigation capability?” The ability to traverse the Lunar surface, for long distances, greatly enables long-term sustained human operations on the Moon, and significantly enhances the productivity of humans in short-duration missions. This is due in part to the euphemism “Many hands make light work”, as well as the consideration that autonomous systems can lay much of the tedious, dangerous, or uncertain groundwork prior to sustained human presence on the Lunar surface. Tunneling, building construction, resource prospecting, resource mining, resource transportation, resource logistics, as well as “claim staking” are all hinged on the ability to place autonomous systems where you want to, when you want to, and doing the activities prescribed. One of the most significant challenges in this research area is the simple cost of field tests on the Lunar surface. The Earth to Moon signal delay is not significant enough to force autonomous or semi-autonomous control, and even when it is semi-autonomous, the “human-in-the-loop” remains significant.1 Tests are being conducted here on Earth to simulate this environment in real-world real-time characteristics, including terrestrial traverses of deserts, and the OffWorld Gym, which have both significantly enhanced our understanding of the challenges of developing and testing these models. Fully autonomous surface navigation on Earth, in confined laboratories and test tracks is difficult enough, and typically has a hard time learning hard lessons. These are all primary concerns, but systems and vehicle engineering required to ensure the vehicle remains functional in the Lunar environment, and has the logistics, failsafe’s, and redundancy built in to ensure mission success is also critical to the effective development and testing of this technology.
As mentioned, several terrestrial-bound experiments have begun to push the boundaries of what is possible via autonomous investigation and travel, notably the Life in the Atacama rover, which has prospected for biological factors in one of the most inhospitable, and surprisingly Mars-like places on Earth, the Atacama Desert in Chile.2 Enabled by solar power, near constant in this cloudless and polar region, the Zoe rover drilled and collected soil samples to determine the concentration and type of microbes living beneath the surface of top soil. In addition to samples, Zoe also took various Infrared Spectrometer images of the surface as it passed over, allowing for a certain amount of organic and non-organic compound identification or classification. While biological evidence was important, in the scope of this short report, we will instead focus on its navigation and autonomous capabilities. In total, 55 km of autonomous rover traverse was conducted, with 10 of 272 commanded traverse being over 1 km in full autonomy. Zoe navigated herself, determined pathways, avoided obstacles, and ensured mission success largely on her own. While the Atacama desert does in fact resemble the Martian surface more than the Lunar surface (drier, rockier), autonomous navigation and traverse in such an austere and radiative environment, controlled via satellite, certainly shows advancement towards a challenging Lunar environment.
While NASA does not define what they mean by “long distances”, it could be reasonable to assume that several dozen kilometers, in a regularly scheduled patter of movement, might not be unreasonable in the preparation for a Lunar research or habitat station. Thus, only exceeding 2 km once in its time in operation, falls short of any type of operational parallel. However, if this success can be replicated via cheaper, smaller, and more intellectually agile rovers, such as those of OffWorld3, then perhaps this astounding success can compound into a workable, adaptable, and generalized autonomous navigation capability preferable for the harsh Lunar environment.