Yutu

CNSA / China Lunar Exploration Program [14].

Overview

Yutu is the six-wheeled rover carried by the Chang’e-3 lander, the second phase of the Chinese Lunar Exploration Program and China’s first soft landing on the Moon [8], [1]. Chang’e-3 was launched on a Long March 3B on 2 December 2013 and landed on 14 December 2013; the lander and rover separated and imaged one another on 15 December 2013 [13]. The rover carried four instruments: a Panoramic Camera, a Visible and Near-Infrared Imaging Spectrometer, an Active Particle-induced X-ray Spectrometer, and a Lunar Penetrating Radar. It drove 114.8 m and stopped moving at the end of its second lunar day [5], [6].

Specifications

Parameter	Value	Source
Wheels	six, three per side, front, middle and rear	[12]
Wheel radius	0.13 m
Wheel width	0.16 m
Wheel surface	grousers in an alternating distribution pattern
Suspension	two-arm linkage per wheel to the chassis
Total mass used in the published multibody model	137.0 kg
Instruments	4	[8]
Dead-reckoning sensors	wheel odometer and IMU	[10]
Localization sensors	stereo navigation cameras

No CNSA page gives the rover mass, dimensions, rated speed, gradeability or obstacle height. The 137.0 kg figure is the total mass adopted in the multibody dynamics model of the vehicle, not an agency specification [12]. The combined dry mass of the Chang’e-3 lander and rover is 1.22 t [6].

Mission profile

Parameter	Value	Source
Launch	Long March 3B, 2 December 2013	[8]
Landing	14 December 2013
Landing region	Sinus Iridum region, northern Mare Imbrium, on the rim of Zi Wei crater	[6], [3]
Lander and rover mutual imaging	15 December 2013	[8]
Surface operations start	15 December 2013	[6]
Distance driven, cross-site visual localization	114.8 m
Distance driven, DOM matching	111.2 m
Farthest VNIS target from the lander	about 40 m	[3]
Mobile period	23 December 2013 to 14 January 2014 for VNIS measurements

Sinus Iridum was selected against combined scientific and engineering constraints, among them the requirement that the landing site lie within the geographic area visible to the tracking stations; Chang’e-3 used two ground stations [6]. The rover made four VNIS soil measurements, at sites 5, 6, 7 and 8, during the period it was mobile [3].

Mobility

The vehicle is a chassis, a suspension system and six wheels, three per side in front, middle and rear positions, each wheel connected to the main body through a two-arm suspension linkage [12]. Wheel radius is 0.13 m and width 0.16 m, with grousers laid out in an alternating distribution pattern rather than uniformly spaced.

Published mobility analysis of the configuration is simulation rather than flight data. A continuum representation model solved by smoothed particle hydrodynamics, run in Chrono against lunar soil parameter sets of 1627 to 1839 kg per cubic meter bulk density and 37.8 to 47.8 degree friction angle, gives tractive force, wheel torque, wheel sinkage and traction slope over slip ratios from 0 to 0.8 at 0.2 m/s translational velocity [12]. Both tractive force and the amplitude of its oscillation rise with soil friction angle and bulk density, and full-vehicle oscillation amplitudes are smaller than single-wheel ones because the suspension adds degrees of freedom. Single-wheel tests in that work loaded the wheel to 11.415 kg of wheel mass plus 11.415 kg on the axis, 22.83 kg total, under lunar gravity [12]. A wheel-geometry sweep over slopes to 17 degrees run in the same model shows that a larger rim diameter or a wider wheel both reduce the slip needed to climb a given slope, and that doubling grouser count from 20 to 40 changes almost nothing [12].

The only in-situ estimate of soil strength at the Chang’e-3 site comes from later analysis of the follow-on Chang’e-4 rover: track-imagery-derived cohesive modulus at the CE-3 site (kD^0.39 = 0.55 to 1.19 N/cm^2.8) came out markedly weaker than at the CE-4 farside site and weaker than the Apollo 15 value, attributed to lesser regolith maturity at CE-3 [9]. A separate terramechanical fit to a Yutu-2 slip event gives farside regolith a bearing capacity near 4 kPa, above the 2 to 3 kPa measured along the Lunokhod 1 traverse, a comparison point for the same wheel class as Yutu’s [4].

Mobility ended after the second lunar day. Wang and colleagues record that the rover traveled about 114 m before a technical problem, while the lander continued to work for a long period afterwards [6]. VNIS soil measurements were taken between 23 December 2013 and 14 January 2014, described as the period during which Yutu was mobile [3]. The rover continued to return data after that date but did not add odometry.

Power and energy

Both solar wings are deployed to either side of the chassis. The Active Particle-induced X-ray Spectrometer sensor head carries a radioisotope heater unit to keep it warm [2], which places the vehicle in the class of Chinese lunar surface vehicles that use radioisotope heating rather than electrical heating for lunar night survival. No generation capacity or battery capacity is published on the CNSA pages or in the literature retrieved for this entry.

Thermal

The published thermal detail is instrument level. The APXS sensor head is warmed by a radioisotope heater unit mounted with it at the end of the robotic arm [2]. Survival of the lunar night required stowing the mast and folding the solar wings over the warmed electronics compartment. The survival temperature limits are not published.

Compute and avionics

No processor part, memory size or radiation tolerance approach for Yutu is published. The main body hosts the scientific instruments, the power system, the onboard computer and navigation system and the cameras [12]. The onboard sensing that is documented is the dead reckoning chain, a wheel odometer and an inertial measurement unit, whose output provided real-time rover position along the traverse [10].

Autonomy

Yutu was teleoperated rather than autonomous. The rover was driven from the Beijing Aerospace Control Center, with the Planetary Remote Sensing team at the Institute of Remote Sensing and Digital Earth supplying the geospatial products [3]. Dead reckoning from the wheel odometer and IMU accumulates error from wheel slip and IMU drift, so two visual methods were applied on top of it: cross-site visual localization, based on ASIFT feature matching and bundle adjustment of navigation camera images taken at adjacent waypoints, and DOM matching, based on matching digital orthophoto maps generated from navigation camera stereo pairs against basemaps built from orbital or descent imagery. Cross-site visual localization ran in near real time, a couple of minutes after image downlink, at every waypoint, and so was the method available to the drive cycle; DOM matching was applied only where distinguishing features such as small craters or large rocks were present in the local area, and returned the most accurate result.

Cross-site visual localization matched tie points between adjacent-waypoint stereo pairs with ASIFT inside search regions predicted from dead reckoning, rejected outliers by an inter-point distance test, and closed the pose with a two-site bundle adjustment. Measured against indoor truth in a simulated lunar field, its error ran 1 to 4 percent of traverse with a 3.3 percent mean, against 15 percent for dead reckoning alone; in flight the pipeline ran on a 32-core ground workstation in under 2 minutes per site including data transfer [10].

The two methods disagree by 3.6 m over the full traverse, 114.8 m against 111.2 m, and both are more accurate than dead reckoning [10].

Communications

The rover communicated through the Chang’e-3 lander and the ground stations of the Chinese deep space network. Chang’e-3 mission design constrained the landing site to lie within the area visible from the tracking stations, of which two were used [6]. No rover link frequencies or data rates are published.

Payload and instruments

Instrument	Function	Source
Panoramic Camera	close-up surface imaging	[1], [8]
Visible and Near-Infrared Imaging Spectrometer (VNIS)	mineralogy from reflectance spectra	[1], [3]
Active Particle-induced X-ray Spectrometer (APXS)	elemental abundances by X-ray fluorescence	[1], [2]
Lunar Penetrating Radar (LPR)	subsurface structure	[1], [8]

VNIS combines a visible and near-infrared imaging spectrometer covering 450 to 950 nm with a shortwave infrared spectrometer covering 900 to 2395 nm and a white calibration panel; the visible channel has 100 channels and the shortwave channel 300, both at 5 nm sampling interval [3]. VNIS spectra of freshly excavated soil against the surrounding weathered material at the rover’s location measured the optical effect of lunar space weathering directly, the same maturity-driven change in reflectance that any rover-borne spectrometer or vision system operating on old regolith has to account for [7].

APXS parameters [2]:

Parameter	Value
Detector	silicon drift detector
Effective area	7 square millimeters
Energy resolution	better than 140 eV at 5.9 keV
Energy range	0.5 to 20 keV
Detection distance	10 to 30 mm
Excitation sources	4 x 70 mCi 55Fe and 4 x 5 mCi 109Cd
Sensor head mass	754 g

The APXS sensor head is carried on a robotic arm at the front of the rover, which places it against a target on the lunar surface for a measurement [2]. Ground verification against blind samples established that major-element abundances are recovered with relative deviations below 15 weight percent at 30 mm detection distance and 30 minutes of acquisition, with detection limit rising as standoff grows and falling as integration time grows, leading to the operational recommendation to keep the sensor head under 50 mm from the target [5].

The Lunar Penetrating Radar was the first ground-penetrating radar operated on the lunar surface, and its dipole antennas are the booms visible fore and aft of the chassis [1].

Modes of operation

The documented cycle is waypoint-based: drive to a waypoint, acquire navigation camera stereo pairs, downlink, localize, plan the next segment [10]. Instrument measurements were made at stops, with the APXS arm deployed to the surface [2] and VNIS pointed at a soil target [3]. Survival of each lunar night required a hibernation configuration with the mast stowed and the solar wings folded.

Ground operations

Rover teleoperation was performed at the Beijing Aerospace Control Center by a rover teleoperation team, working with the Planetary Remote Sensing team of the State Key Laboratory of Remote Sensing Science at the Institute of Remote Sensing and Digital Earth, Chinese Academy of Sciences [3]. The geospatial products supporting operations were landing site mapping, topographic analysis, hazard detection, path planning, rover localization and navigation, built from orbital data, descent imagery and rover navigation camera stereo pairs.

Technologies developed

The localization chain built for Yutu was carried directly into Chang’e-4. Cross-site visual localization and DOM matching, developed and applied at Beijing Aerospace Control Center for Chang’e-3, were the same techniques used to localize Yutu-2, which traveled 190.66 m by odometer and 186.66 m by visual localization through its fifth lunar day, and 345.059 m as of 3 December 2019 [3], [11]. Chang’e-4 was built as the backup spacecraft to Chang’e-3 and reused its landing site selection method [6]. In Yutu-2 operations the two channels were kept as a cross-check rather than a handoff: onboard dead reckoning and ground-based visual positioning agreed to the centimeter or decimeter level over the up to 10 m between neighboring waypoints, and diverged by more than 1 m only where the rover had slipped badly, which is when the visual correction mattered [11].

The Lunar Penetrating Radar established subsurface sounding from a rover as a lunar technique [1], and the APXS design, a 754 g silicon drift detector head on a deployable arm with a radioisotope heater unit and 55Fe and 109Cd excitation sources, was the basis for the instrument carried on Yutu-2 [2].