A 2-year locomotive exploration and scientific investigation of the lunar farside by the Yutu-2 rover

Abstract

The lunar nearside has been investigated by many uncrewed and crewed missions, but the farside of the Moon remains poorly known. Lunar farside exploration is challenging because maneuvering rovers with efficient locomotion in harsh extraterrestrial environment is necessary to explore geological characteristics of scientific interest. Chang’E-4 mission successfully targeted the Moon’s farside and deployed a teleoperated rover (Yutu-2) to explore inside the Von Kármán crater, conveying rich information regarding regolith, craters, and rocks. Here, we report mobile exploration on the lunar farside with Yutu-2 over the initial 2 years. During its journey, Yutu-2 has experienced varying degrees of mild slip and skid, indicating that the terrain is relatively flat at large scales but scattered with local gentle slopes. Cloddy soil sticking on its wheels implies a greater cohesion of the lunar soil than encountered at other lunar landing sites. Further identification results indicate that the regolith resembles dry sand and sandy loam on Earth in bearing properties, demonstrating greater bearing strength than that identified during the Apollo missions. In sharp contrast to the sparsity of rocks along the traverse route, small fresh craters with unilateral moldable ejecta are abundant, and some of them contain high-reflectance materials at the bottom, suggestive of secondary impact events. These findings hint at notable differences in the surface geology between the lunar farside and nearside. Experience gained with Yutu-2 improves the understanding of the farside of the Moon, which, in return, may lead to locomotion with improved efficiency and larger range.

WayFAST: Navigation With Predictive Traversability in the Field

“…We use a nonlinear moving horizon estimator (NMHE) to generate traversability labels to train our prediction network. The NMHE uses the system model shown in (1) and synchronously runs with GNSS on the robot. We define μ and ν shown in (1) as the traversability coefficient.…”

“…For example, tall grass could be detected by LiDAR or tactile sensors as geometric obstacles, when in fact, these are obstacles that the robot could traverse through. Moreover, in addition to predicting obstacle free paths, the traversability system should also predict expected traction coefficient to ensure successful autonomous navigation [1], [2].…”

WayFAST: Navigation with Predictive Traversability in the Field

“…We use a nonlinear moving horizon estimator (NMHE) to generate traversability labels to train our prediction network. The NMHE uses the system model shown in (1) and synchronously runs with GNSS on the robot. We define µ and ν shown in 1 as the traversability coefficient.…”

“…For example, tall grass could be detected by LiDAR or tactile sensors as geometric obstacles, when in fact, these are obstacles that the robot could traverse through. Moreover, in addition to predicting obstacle free paths, the traversability system should also predict expected traction coefficient to ensure successful autonomous navigation [1], [2]. Learning-based methods to determine traversability based on navigation experiences could lead to more robust navigation by avoiding heuristics [3], [4], [5], [6], [7].…”

Direct photoconversion of CO2 to carbon nanotubes for oxygen recycling in extraterrestrial exploration

“…As a philosophically and scientifically important pursuit, space exploration would provide humanity clues for the origin of "us" and life beyond Earth, and would also offer the way to search habitable planets 1,2,3 . The National Aeronautics and Space Administration (NASA), Russian Federal Space Agency (RKA), European Space Agency (ESA), and China National Space Administration (CNSA) have drawn up plans for future manned near-Earth and deep space exploration programs.…”