Localization and Surface Characterization by Zhurong Mars Rover at Utopia Planitia Localization and Surface Characterization by Zhurong Mars Rover at Utopia Planitia Extended Data Table Daily scientific exploration events. Zhurong carried out routine navigation and detection using NaTeCam, Mars rover penetrating radar (RoPeR), and Mars climate station (MCS) on most sols. Scientific payloads 2 like multispectral camera (MSCam), Mars surface composition detector (MarsSCoDe), and Mars rover magnetometer (RoMAG) used for a specific target detection, like rocks and sand The a day, to at the of


 China’s first Mars rover, Zhurong, has successfully touched down on the southern Utopia Planitia of Mars at 109.925° E, 25.066° N, and since performed cooperative multiscale investigations with the Tianwen-1 orbiter. Here we present primary localization and surface characterization results based on complementary data of the first 60 sols. The Zhurong rover has traversed 450.9 m southwards over a flat surface with mild wheel slippage (less than 0.2 in slip ratio). The encountered crescent-shaped sand dune indicates a NE-SW local wind direction, consistent with larger-range remote-sensing observations. Soil parameter analysis based on terramechanics indicates that the topsoil has high bearing strength and cohesion, and its equivalent stiffness and internal friction angle are ~1390-5872 kPa∙m-n and ~21°-34° respectively. Rocks observed strewn with dense pits, or showing layered and flaky structures, are presumed to be involved in physical weathering like severe wind erosion and potential chemical weathering processes. These preliminary observations suggest great potential of in-situ investigations by the scientific payload suite of the Zhurong rover in obtaining new clues of the region’s aeolian and aqueous history. Cooperative investigations using the related payloads on both the rover and the obiter could peek into the habitability evolution of the northern lowlands on Mars.


China's first Mars rover, Zhurong, has successfully touched down on the southern Utopia
) and carrying six scientific payloads 2 -7 , becomes one of the most 20 powerful rovers to date that ever landed on the Northern Plain of Mars. Zhurong has been roving and conducting in-situ investigations for more than the designed 90-sols primary mission, collaborating with the orbiter to collect complementary scientific data at different perception scales and precisions for five primary scientific objectives 7 . Scientific data of ~7 Gbit has been relayed back via the Tianwen-1 orbiter 3 and is currently under calibration. Here we present preliminary results in terms of the localization and surface characterization of the landing site based on the first 60-sols cooperative observations from the Zhurong rover and the Tianwen-1 orbiter.
The landing site location was refined to 109.925° E, 25.066° N using the descent image sequence and 463 m·pixel -1 ). About 99% of the area consist of slope less than 2.3°, and relatively large slopes are primarily distributed at the rim of impact craters (Fig. 1c). Therefore, the area is flat and may facilitate the 15 long-distance rovering and exploration of the Zhurong rover.
Zhurong has traveled 450.9 m due south during the first 60 sols (Fig. 2a), with waypoints calculated by cross-site visual localization 10 , and conducted several in-situ investigations (Extended Data Table 2). An off-line dynamic locomotion simulation 11 (Extended Data Fig. 3) implemented on the Rover Simulation based on Terramechanics Dynamics (RoSTDyn) 12 is used to evaluate and verify the safety and efficiency of 20 the planned path. Compared with the visual localization postures, the simulations of planned traverses deviated by ≤ 2%, and the average azimuth error was ≤ 1°. Due to the inevitable wheel slippage, the planned destination was not always accurately reached, particularly when traversing through drifting soils.
The wheel slip ratio (denoted by s) over each continuous steady traverse is calculated using the Guidance, 4 Navigation and Control System (GNC)-derived rover velocity and the wheel angular velocity derived from the encoder (see Methods). The distribution of the wheel slip ratio along the route (Fig. 2b) shows that the Zhurong rover primarily operates under mild wheel slippage (s < 0.2). The average slip ratio of a steady climb of ~0.55° was 0.056, consistent with the elevation increase of 4.34 m during the 60-sol journey (Fig.   2c). The largest average slip ratio of 0.22 occurred on sol 55 while Zhurong climbed a slope of ~1.14°. 5 Occasionally, the rover suffers wheel skids (s < 0) on localized down slopes, e.g., traversing through a small down-slope area of ~-3° on sol 34 before approaching a crescent-shaped dune. Wheel tracks in periodic textures (Extended Data Fig. 4a) left by the Zhurong rover also provide clues on the severity of wheel slippage in a more detailed view. By extracting the wheel track unit (see Methods), the average wheel slip ratio on legible parts of the wheel track (being created after Zhurong traversed from waypoints X 10 to A; Extended Data Fig. 4b) was calculated to be 0.05, consistent with telemetry-derived results.
The first 360° panorama is stitched from twelve images taken by the Navigation and Terrain Camera (NaTeCam) (Fig. 3a). The panorama shows an overview of gentle topography at the landing site, with major surface geological features including aeolian dunes, small craters, and rocks. One distinct feature is that the distribution of several aeolian dunes with ripple-like structures, which can be reached and 15 investigated by the rover. Most dunes in this area are bright-toned and crescent-shaped, with windward slopes (the protruding side) oriented approximately north-eastwards, indicating a NE-SW local wind direction (Extended Data Fig. 2b). The first dune encountered by the Zhrong rover on sol 50 is 40 m in length, 8 m in width, and 0.6 m in height (Fig. 3b), with sands in two different colors covered the dune surface. Several seagull-shaped deposits are usually formed from two merged dunes, and one approximate 20 to a line as shown in the enlarged view (Extended Data Fig. 2b). Further scientific observations from the orbit are needed to determine whether these dunes are currently active. A fresh mini crater (~0.95 m in diameter) beneath the lander created by the landing plume was observed in the NaTeCam image, providing clues on a shallow subsurface layer (Fig. 3c). Gravels (1-4 cm in diameter), splashed out around the lander 5 struts or settled on the crater rim, exhibit a dark-brown tone and in sharp contrast to the semi-buried ones by the dust.
The mechanical properties of the surface soils (Fig. 3d) are estimated using the lugged wheels of Zhurong as test devices. Using stereo images taken by the rear Hazard-avoidance Camera (HazCam), the soil surface deformation by lugged wheels can be reconstructed and the wheel sinkage can be estimated by 5 extracting the wheel track profile, as it was done for the Chang'E-3 mission 13 . In legible parts of the wheel tracks (Extended Data Fig. 4d), the equivalent wheel sinkage (the sum of the contribution of the lug and the drum-shaped wheel) was approximated to be 10 mm. However, most wheel tracks are incomplete without well-trimmed contours (Extended Data Fig. 4e). The wheel tracks are formed in a way that the 5-mm-high wheel lugs almost completely immerse into the soil, with the wheel rim several millimetres above the 10 surface (as observed in the images on sol 12 taken by a wireless camera deployed on the ground; Extended Data Fig. 4c). Therefore, the equivalent wheel sinkage was estimated to be about 5 mm. Wheel sinkage is sometimes interrupted by protruding gravels on the surface , leading to smaller values (about 2 mm) or going larger to approximately 15 mm when the wheel rim sinks below the surface (Extended Data Fig. 4f).
Compared with Yutu or Yutu-2 rovers of the lunar exploration missions, the load on each wheel of 15 Zhurong (~148.8 N) is much larger than that on the wheels of lunar rovers (~36.5 N). However, the corresponding wheel sinkage during the Zhurong traverses was not more significant, indicating a greater bearing strength of the Martian soil than that of the lunar regolith. According to the terramechanics model 14 for the lugged wheels, characteristic curves of the sinkage exponent (denoted by n) and the equivalent stiffness (denoted by Ks) were predicted under different sinkage conditions (Fig. 4a). We set the upper 20 bound of n to 1.0 (i.e., the typical value of lunar regolith 15 ). As n increases, the corresponding soil soften and results in a sharp increase in wheel sinkage. Also, the lower bound of n is set to 0.7 to maintain Ks within the reasonable range because the Ks decreases with a decresing n value. Considering soil characteristics within curves corresponding to a wheel sinkage of 2-5 mm and representative terrestrial soil 6 types, the equivalent stiffness Ks was estimated to be 1390-5872 kPa•m -n , shown as the rectangular region in Fig. 4a.
Using the wheel-terrain interaction model, the characteristic shearing curves under different driving torques (4.0-8.0 N•m) were predicted (Fig. 4b). Transformed from the wheel driving motor current in moving states, the average wheel driving torque during each continuous traverse during sols 23-34 ranged Zhurong has encountered are relatively small (< 10 m in diameter) (Extended Data Fig. 5c, d) in a depression-shape, surrounded by dark rocks (probably ejecta). Aeolian deposits within the craters may indicate that the craters may be subject to long-term weathering.
The surface is primarily free of large boulders but is scattered with small rocks and clasts bearing distinct features (Extended Data Fig. 6). Most rocks are fine-grained in texture and angular to subangular, 5 with low roundness in morphology. Some rocks with pitted surfaces show similar morphology to igneous rocks as observed in previous missions (e.g., by the Viking-1 lander and the Spirit rover) 24,25 , which are proposed to form via brine-related dissolution processes under cold environments 25,26 . Similar rock features may provide insights into the climate and geological processes of the Tianwen-1 site but require detailed investigations by the scientific payloads of the rover. Some rock chucks also show flaky texture, similar to 10 the rock targets observed at Gusev Plain, such as "Mimi" 27 . The flake textures might be related to aqueous alterations that dew water goes through insolation cracking to flake the rocks, and brine and salt may work to cement these flakes 28 . In addition, rocks with grooves and etchings on the windward sides are ubiquitous on the landing site, usually interpreted as ventifacts resulting from intense wind erosion with aeolian particles 29-31 . Therefore, the rock textures observed at the site so far may indicate both presences of 15 physical weathering (e.g., impact sputtering, wind erosion, and potential freeze-thaw weathering) and aqueous interactions, involving salt and brine activities. These rock and soil targets provide excellent opportunities to peek into the aqueous history and climate evolution of the northern lowlands and shed light on the habitability evolution of Mars.  Only the sols when Zhurong moves in high-efficiency mode (from sol 23 on) are plotted. Each box represents the distribution of wheel slip ratios over continuous traverses experienced in that sol. Traveling distances on sols 42-48 are short, causing relatively large locomotion measurement errors and resulting in the slip ratio in disagreement with the elevation trend. c, The elevation profile along the traversed path 10 during the first 60 sols. The blue dots represent the rover elevation of starting waypoints on each sol, while the yellow dots represent the elevation within the waypoints. The rover elevation of waypoint X (the initial waypoint on the surface) is taken as the baseline (elevation of 0), and the elevation varies from 0 (on sol 10) to 4.34 (on sol 60) m. 15 Fig. 3. The geological features at the landing site. a, The first 360° panorama of the Tianwen-1 landing site taken by the NaTeCam. The image is stitched by 12 images taken at 30° intervals by the NaTeCam on sol 6. The Zhurong rover is heading southwards, and the parachute and backshell are located a few hundred meters to the southwest of the lander. Two jet-wasted traces reveal darker materials below the topsoil. On 5 the southeast of the lander, there are several bright sand dunes, a small crater surrounded by several darkcolored rocks, probably ejected from the crater. b, The first aeolian dune that Zhurong rover has encountered. The windward slope is relatively gentler than the leeward one. The image was taken on sol 50. c, A mini crater (~0.95 m in diameter) right underneath the lander formed by the thrust engine plume of the Tianwen-1 probe during landing. The crater reveals two distinct rock fragments. The one with a dark-10 brown tone contrasts sharply with the bright-tone clasts semi-buried in the soils and dust. d, Soil surface scattered with bright-toned clasts (green arrows) and small-sized dark-toned rocks (yellow arrows).

Instruments and data description
The Zhurong rover is a six-wheeled solar-powered robot with active suspension. Benefited from its active suspension structure 32 , the rover is not only able to work in basic wheeled movement modes but also capable of novel gaits, such as a crab gait for cross-walking, a creeping gait for better slope climbing 5 capability, a wheel uplift gait to escape from wheels being stuck 33 , and a body uplift/settlement gait. The strong terrain adaptability and resilient fault recovery capability allow the rover to access dangerous but scientifically beneficial regions, such as sand dunes and crater rims, leading to significant scientific discoveries. The lugged wheels on the Zhurong rover are used as devices for soil mechanical parameter analysis based on terramechanics. Considering the gravity on Mars, the vertical load on each wheel is 10 estimated to be 148.8 N under a quasi-static state, based on the rover's configuration and mass distribution.
The parameters of the drum-shaped wheel are in the Extended Data Table 1.
The data used in this study includes images from a NaTeCam, two HazCams, and a wireless camera, alongside locomotion data from the onboard inertial measurement unit and wheel encoders. The locomotion data used in this study included the rover's position (x, y and z) and posture (roll, pitch and yaw) at the landing site local (LSL) coordinate frame, the rover's velocities along three axes in its coordinate frame, and the angular velocities of each of the wheels. The rover's positions and postures are 5 derived from the inertial measurement unit onboard, and the wheels' angular velocities are derived from the angle recorded by wheel encoders. The LSL coordinate frame is a north-east-down right-handed coordinate system with its origin at the first waypoint. Its z-axis points down in the local normal direction, the x-axis points to the north pole, and the y-axis is orthogonal to the x-and z-axes. The rover coordinate frame is an east-north-up right-handed local system, whose origin is at the rover's centre, its x-axis points to the 10 forward direction of the rover, its z-axis points in the local normal direction, and its y-axis is orthogonal to the x-and z-axes. These data were recorded onboard at a higher frequency, but the data transmitted to Earth are only at a frequency of 1/15 Hz due to the communication channel restriction, with both recorded and received timestamps. 15 The slip ratio s of a lugged wheel at each moment t is defined as follows:

Wheel slip ratio estimation based on telemetry data
where ω(t) is the angular velocity function, rs is the shearing radius function, and v(t) is the linear velocity function. The wheel angular velocity is recorded by wheel encoders, and its linear velocity is derived based on the rover's linear velocity (recorded by the onboard inertia measurement unit) and the curvature of the 17 where r is wheel radius, h is the lug height and s λ ( s 1 0 λ ≤≤ ) is the lug coefficient determined by the number of lugs and the internal friction angle of the soil. As Zhurong's wheels are drum-shaped, its radius r in equation (2) is set as a value between the largest radius and the smallest radius. Here, it is set as the mean of wheel's largest radius and its smallest radius. Besides, Zhurong's wheels are with evenly arranged 5mmhigh lugs, thus the value of s λ is approximately 0.5 according to the experiment results in 34. 5 A slip ratio s value larger than zero indicates that the wheel has slipped; an s equal to 0 indicates that the wheel has rolled without slipping or skidding; and an s less than 0 indicates that the wheel has skidded, with |s| being the values of the skid ratio. Generally, a driving wheel slips when moving on flat terrain or climbing up a slope, and may skid when moving down a slope 33 .
Slip ratio estimation based on wheel imprint 10 The slip ratio of a Zhurong rover wheel can be estimated from the imprint of the wheel. ( ) ( ) where n is the number of lugs on the wheel. Therefore, s can be estimated by measuring p x ∆ .
The width of the trace unit p x ∆ is estimated using NaTeCam or rear HazCam images (Extended Data   Fig. 4b, c). An image processing method was introduced to extract p x ∆ with two key steps: view correction 20 and trace feature extraction. The original image is first corrected with the camera extrinsic matrix to a topview image. Then, each trace imprint unit is manually divided into independent areas, and trace imprint 18 information, such as the trace unit width, is acquired. The wheel slip ratio is then estimated from the trace imprint using equation (3). When a series of continuous trace units has almost the same width, the average trace unit widths are calculated to reduce the measurement error, and the average wheel slip ratio is deduced to represent the wheel slippage experienced by Zhurong. Since only whole imprints of the rear wheels are left behind the rover while the front and the middle wheel imprints are typically overlapped, the 5 slip ratio of the rear wheels was estimated to represent the rear wheel slippage at the respective moment.

Soil parameter analysis
Zhurong's wheels were used as devices to analyze soil parameters based on wheel-soil interaction. For a lugged rover wheel moving on the soil with an angular velocity ω, the wheel is applied with a vertical load W and a resistance force fDP from the vehicle suspension, as well as a driving torque T at the wheel 10 rotational axis by an actuator. The terrain interacts with the wheel circumference in the contact region, which corresponds to the angle divided into two parts: the entrance angle θ1 from the vertical at which the wheel first makes contact with the soil, and the exit angle θ2 from the vertical, at which the wheel loses contact with the soil. In the wheel-terrain interaction region (θ1+θ2), the continuous normal stress σ to support the wheel, and the shearing stress τ due to the relative movement are exerted on the wheel surface, 15 as shown in Extended Data Fig. 7a. The point of maximum stress is denoted as θm, according to which the stress region is divided into a forward part (σ1, τ1), corresponding to the angle from θ1 to θm, and a rear part (σ2, τ2), corresponding to the angle from θm to θ2.
The soil bearing parameters are closely related to the wheel-soil interaction in the normal direction.
where b is the width of the wheel, rs is the equivalent wheel radius calculated on equation (2), c k is the cohesive modulus of the soil, k ϕ is the frictional modulus of the soil, and n is a soil sinkage exponent that can be represented by a linear function of slip ratio s as follows: The entrance angle θ1 is a function of wheel sinkage z as: 10 and the exit angle θ2 is computed as: where c3 is a coefficient of the wheel-terrain interaction angle, generally assumed to be 0 38 . The exit angle θ2 is used to compute the stress caused by the rebound of the soil (stress integration from θm to θ2), which is usually negligibly small and is taken as zero in the calculation. 15 The maximum stress angle θm is computed as: where c1 and c2 are coefficients of the wheel-terrain interaction angle. In the calculation, c1 and c2 are set to 0.5 and 0, respectively, because it is reasonable to assume that the angular location of maximum stress θm occurs midway between θ1 and θ2 39 . 20 20 The tangential stress τ(θ) is computed as where c is the cohesion of the soil, ϕ is the internal friction angle, k is the shearing deformation modulus, rs is the equivalent shearing radius calculated on equation (2), and 1 θ ′ is the equivalent entrance angle. With the lug effect, the shearing stress occurs on the surface of the soil sticking to the wheel circumference due 5 to the wheel lugs instead of the wheel outer cylinder surface. The equivalent entrance angle 1 θ′ of a lugged wheel 14 is computed as When θ approaches m θ , the corresponding normal stress and tangential stress approach their maximum as When the wheel is in a quasi-static state, the effect of the distributed stress (normal stress σ and tangential stress τ) can be simplified to the normal force FN, drawbar pull FDP, and driving resistance torque MR by integrating along with the wheel-terrain interaction area, which are balanced with the wheel load W, resistance force fDP, and driving torque T, respectively. For the wheels of Zhurong moving almost on flat 15 terrain with a maximum speed of 200 m/h, the quasi-static condition is valid because the dynamic effects are negligible at low speeds. Therefore, the force/torque balance equations for Zhurong's lugged wheels can be expressed as follows: According to Shibly et al. 39 , the wheel-terrain interaction stress can be linearized as follows: Because the entrance angle is usually not large, and when θ approaches θ1, the corresponding normal 5 stress and tangential stress all approach zero, Ding 40 proposed that the product of cosθ and stress can also ( )  Regarding the analysis of the soil shearing parameters, we find equations (12) and (21)  analysis of the rover with knowledge of the rover configuration and mass distribution. The two key wheel motion state indicators, the slip ratio s and sinkage z, can be computed using vision-based techniques or kinematic analysis of the rover suspension. Based on the given parameters, for wheel sinkages of 2, 5, 10, and 15 mm, the curves of the equivalent stiffness modulus (denoted by Ks) and the sinkage exponent n were plotted (Fig. 4a). The motor current of the driving wheel varied from 0.17 A to 0.28 A with an average of Eight continuous track units are extracted on the orthograph for local wheel slip ratio calculation. The image was taken by the NaTeCam on sol 11. c, Near-Side view of the wheel-terrain interaction. Only wheel 5 lugs are submerged into the soil and the wheel rims of these three right wheels are millimetres above the surface. Some soil adhered to the wheel surface or on the groove bordering the lugs. This image was taken by the wireless camera on sol 12 when the rover was retreating. d, e, f, show wheel tracks and the associated wheel sinkage. Images were taken by the rear HazCam. d, A part of legible wheel track with well-trimmed contours. The wheel sinkage is estimated to be about 10 mm. e, Most common form of wheel 10 tracks without well-trimmed contours. Its wheel sinkage is estimated to be ~5 mm. f, shows both wheel tracks interrupted by gravels and wheel tracks with rim sinking below the surface. The wheel sinkage of the wheel tracks interrupted by gravels is estimated to be ~2 mm. The wheel sinkage of wheel tracks formed by the wheel rim sinking below the surface is estimated to be ~15 mm.  Six wheels on the Zhurong rover can both driving and steering. The wheel's largest radius (145 mm) is in the middle crosssection and the smallest radius (135 mm) is on the two end faces at both sides. The 20 wheel lugs are evenly attach to the outer edge of the wheel. 5 Extended Data Table 2. Daily scientific exploration events. Zhurong carried out routine navigation and detection using NaTeCam, Mars rover penetrating radar (RoPeR), and Mars climate station (MCS) on most sols. Scientific payloads 2 like multispectral camera (MSCam), Mars surface composition detector (MarsSCoDe), and Mars rover magnetometer (RoMAG) are used for a specific target detection, like rocks and sand dunes. The sol represents a Martian day, corresponding to 24.65 hours at the early stage of the