Sojourner

NASA/JPL. Public domain (NASA / US government work).

Overview

A rover meant to survive a week on Mars had to be commanded, sensed and moved without knowing anything about the ground under it, because no vehicle had done it before and there was no prior surface data to design against. Sojourner rode the Mars Pathfinder lander to Ares Vallis and became the first rover to operate on Mars on 4 July 1997 [1], [2]. Planned for seven sols, it returned data and images for 83 [3].

Specifications

Parameter	Value	Source
Mass	10.6 kg (15.2 kg with lander support equipment)	[3]
Height	about 30 cm	[1]
Wheels	six, rocker-bogie suspension	[4], [5]
Instruments	alpha proton X-ray spectrometer, three cameras	[1]

Mission profile

Parameter	Value	Source
Launch	4 December 1996 UTC, Cape Canaveral	[1], [2]
Landing	4 July 1997, Ares Vallis	[1], [2]
Planned mission	seven sols	[4]
Surface operations duration	83 sols	[3]
Final data transmission from Pathfinder	27 September 1997	[1]
Linear distance traversed	approximately 100 m	[4], [2]
Average traverse per sol	2.7 m	[3]
Commanded rover maneuvers	230	[4]
APXS in-situ elemental analyses	16	[4]
Rover camera images	550	[4], [1]
Environmental measurements	approximately 8.5 million	[4]
Total mission data returned	approximately 2.3 Gbit	[4], [1]

The mission page counts more than 15 chemical analyses of rocks and soil [1], where the rover engineering report gives 16 APXS in-situ analyses [4]. The 83 sols and roughly 104 m of total traverse are far beyond the design life of one week for the rover and one month for the lander, so nearly every figure in the tables below describes an anomalously long-lived technology demonstration rather than typical performance for the design [2].

Pathfinder entered the atmosphere directly, decelerated by heat shield, parachute and retrorockets, and landed inside an airbag cocoon [2]. The project’s own account, filed at about sol 60, gives the entry, descent and landing sequence directly from telemetry: entry deceleration, the altitude of parachute deployment, and the altitudes of rocket ignition and lander separation [6]. Airbag landing is viable at this mass because the atmosphere, at a mean surface density near 0.020 kg/m3, decelerates a light entry vehicle adequately under a parachute alone [7].

Mobility

The rover masses 10.6 kg on six wheels in a rocker-bogie suspension [4]. The linkage’s proportions were set by solving, at each of several candidate obstacle geometries, the coefficient of friction a wheel needs to climb a vertical face, then choosing link lengths to minimize the worst case: six wheels clear an axle-high wall at a friction coefficient near 0.8, where a four-wheeled layout needs 0.78 to 1.4 depending on which end meets the obstacle [5]. Springs were left out deliberately, because an elastic suspension raises the load on a climbing wheel while unloading the others, the opposite of what obstacle climbing needs [5]; the same passive linkage keeps all six wheels loaded on uneven ground without active levelling. The same geometry, scaled up and re-tested on the Exploration Technology Rover precursor to the Mars Exploration Rovers, has been used on every subsequent NASA Mars rover [8].

Wheel surface speed during instrumented operation was 1 cm/s at approximately 1.2 rpm [4], a rate low enough that wheel-soil interaction is quasi-static and inertial effects in the soil do not enter. Wheel dig-in tests give the only direct in-situ measurement of surface bearing behavior from this mission: in softer soil the test wheel sank 0.6 to 0.7 cm in the first two revolutions and reached 0.8 to 1.2 cm after seven, while in harder material it reached 1.5 cm in two revolutions. Ground tests in analog soils produced depths over 3 cm [4]. Separate wheel-torque shear tests at other sites on the same traverse gave friction angles near 34 to 39 degrees for the cloddy material and 26 to 28 degrees for the loose drift deposits, with small cohesion in both [9]. A companion analysis of twelve wheel-spin experiments derived apparent cohesions of a few tenths of a kilopascal by assuming the friction angle equals a photographically measured angle of repose, an assumption its own least-squares fits do not consistently satisfy, so those specific cohesion figures should be read as illustrative rather than tightly bounded [9]. Taken together, the spread reflects the strength range of Martian surface materials, which run from sand at cohesion at or below 1 kPa to indurated soils near 11 kPa [10].

Power and thermal

A solar array on the top surface supplies about 16 W, with non-rechargeable lithium batteries for short night operations [4]. Electronics rode in a warm electronics box lined with silica aerogel, sized with essentially no margin between the predicted flight temperature and the allowable limit beyond the 10 to 15 degC gap to the qualification temperature [11]. Mean surface pressure at the site was 6.7 mbar [4], the thin atmosphere that made that thermal margin necessary in the first place. In flight the computer and power boards held between about -20 and +40 degC each sol, and the only component that misbehaved was the commercial UHF modem, whose crystals drifted with temperature; the fix was operational, timing transmissions and heating cycles around the sol rather than changing the hardware [1]. Postflight correlation found the preflight thermal model had missed ground-reflected solar flux onto the cameras, worth about 20 percent of the insolation, and a near-surface wind sublayer where effective wind speed is under half of what the lander mast measured [1].

Atmospheric opacity at the landing site was 0.5 during surface operations, with suspended particles of approximately 1 µm [4]. That opacity sits at the top of the background range for the aphelion half of the Martian year, so array output over the 83 sols was near its seasonal floor rather than depressed by a storm [7].

Hazard sensing and the Earth link

Hazard detection uses two binocular cameras and a laser ranging device; five laser stripes gave twenty elevation estimates 56 to 81 cm ahead, acquired only while the rover was stationary [4]. Neither sense soils strength, and cohesion across Martian surface types varies from at or below 1 kPa for sand to 11 kPa for indurated soil with no corresponding change in relief [10]. The rover has no direct Earth link; all communication relays through the Pathfinder lander over UHF at a raw rate of 9,600 bps, about 2,000 bps effective, out to a 500 m maximum range [3]. One-way Earth-Mars signal delay during the mission was at least ten minutes, which is what forces onboard hazard handling rather than direct teleoperation.

Autonomy

Commanding was once per sol, so a Go To Waypoint command had to run without oversight for its entire length: the rover drove an approximate straight line, stopped roughly every 7 cm of travel to difference images taken with and without the laser stripes, and dead reckoned position from wheel encoders and a rate gyro, with a measured result, from the only 83-sol record of 1990s planetary rover autonomy that exists, of position error of about 5 to 10 percent of distance traveled, heading drift of roughly 1 to 3 degrees per sol, and an average traverse of 2 to 3 m per sol [3]. Hazard detection kept the vehicle off non-traversable terrain even where dead reckoning alone would not have gotten it to the commanded point, which is the split in performance that the flight record establishes; the team also states it withheld the full hazard-avoidance suite on many sols because a bad traverse risked the mission, so the flight numbers understate rather than bound what the software could do [3]. The dig-in results show why geometric sensing is only a partial safeguard: the same seven wheel revolutions produced 0.8 to 1.2 cm of sinkage in one material and 1.5 cm in two revolutions in another, a difference in bearing behavior no laser or camera resolves [4], [10]. The relay-only link also confined the rover to within its UHF range of the lander [3].

Payload and instruments

The alpha proton X-ray spectrometer was placed against targets by backing the rover into them, returning the first in-situ measurements of Martian rock composition; 16 analyses were completed [4]. Close-range Sojourner imagery of rounded, centimeter-scale objects and of knob-and-socket textures on nearby rocks was read by the rover team as possibly pebbles freed from conglomerate, an interpretation offered as one of several competing origins rather than established by composition or texture measurements taken at the time [9].

Wheel Abrasion Experiment

The WAE is the reason Sojourner produced a quantitative result about the abrasiveness of Martian surface material rather than only about its composition.

Parameter	Value
Location	right center wheel
Substrate	black anodized, dye-impregnated aluminum strips, 25 x 120 x 0.25 mm
Metals	aluminum, nickel, platinum, resistively evaporation-deposited
Film thicknesses on one strip	200, 300, 450, 700, 1000 Å
Brinell hardness of the three metals	Al 16, Pt 64, Ni 100
Sensor	photovoltaic reflectance detector
Typical run	2 revolutions acquiring, 3 not, 2 acquiring; seven total
Readings per sample per revolution	five, central reading taken as peak
Runs conducted	11, totaling 39 wheel revolutions

Values from [4].

The metals bracket a hardness range: material that abrades aluminum but not nickel has a hardness between Brinell 16 and 100 [4]. In flight, aluminum wore most and wore inversely with film thickness, while platinum and nickel wore little. The limits placed on Martian dust are therefore harder than aluminum and softer than nickel.

Two experiment modes were run. Rolling wear accumulated during normal driving. At intervals all wheels except the test wheel were locked to hold the rover stationary while the test wheel alone spun into the regolith, producing wear more severe than rolling and the dig-in depths quoted above.

The experiment returned a second result it was not designed for. Electrostatic charging of the wheels attracted dust onto the WAE wheel, which made reflectance interpretation problematic on many runs [4]. Ground tests of a spare WAE wheel in a simulated Martian atmosphere reproduced the effect directly: the wheel picked up dust and developed a measurable electrostatic charge, the charge fell sharply once the wheel was grounded, and the magnitude of charging grew as grain size shrank [4]. The surface-deposit survey ties the same mechanism to the reflective, low-rimmed rover tracks left in material with grains below about 40 µm, rather than in sand [9]. That the WAE dust adhesion and the track reflectivity point to the same grain-size threshold is what makes electrostatic charging, rather than simple mechanical clogging, the identified cause.

WAE photodetector

Parameter	Value
Detector	three GaAs-on-Ge solar cells, 2 x 4 mm each
Package	1 cm cube, 10 g mass budget
Design / actual operating temperature	-50 C / -30 C
Load resistor	1.26 kΩ at 0.1 percent, against a calculated maximum of 1.275 kΩ
Output voltage limit	250 mV
Peak response	5 pA·cm²/mW at 0 C
Short-circuit current at 40 mW/cm²	196 pA
Isc temperature coefficient	0.067 percent per C
Pointing tolerance	±15 deg
Standoff, sample center to detector face	19.2 mm
Imaged sample area	1.5 x 2.5 cm
Landing load qualification	60 g
Low temperature qualification	-110 C, 1 h dwell
Bakeout	110 C minimum, 50 h
Vacuum test	below 1 torr, 24 h at 40 C

Values from [12].

Detector output feeds the reflectance record from which the metal film wear is read [4]. The load resistor is the design constraint that decides the whole measurement. Photocurrent at maximum illumination is 196 pA, so the resistor must be large enough to develop a readable voltage and small enough that the cell stays near short circuit; 1.26 kΩ against a calculated 1.275 kΩ maximum leaves no margin above, and the 250 mV output ceiling is what sets it [12]. Detector temperature was specified at -50 C and the flight units ran at -30 C, over which range the short-circuit current moves 0.067 percent per degree.

The detector was qualified to 60 g of landing load, to -110 C with a one hour dwell, and through a 110 C bakeout of at least 50 hours plus 24 h below 1 torr at 40 C [12]. Its viewing geometry is fixed at 19.2 mm standoff over a 1.5 by 2.5 cm sample area with a ±15 degree pointing tolerance, so alignment error translates directly into apparent reflectance change and had to be held mechanically rather than calibrated out.

Surface conditions measured

Quantity	Value
Mean atmospheric pressure	6.7 mbar, varying 0.2 to 0.3 mbar daily
Atmospheric opacity	0.5
Suspended dust particle size	approximately 1 µm
Precipitable water vapor	0.01 mm
Maximum daytime / minimum night temperature	-9 C / -76 C
Temperature sensor heights on the lander mast	25, 50, 100 cm

Values from [4].

Winds were light and variable relative to the Viking sites: steady from the south at night, rotating clockwise south to west to north to east through the day. Later compilation across all landed sites puts daytime wind maxima and a clockwise diurnal rotation of wind direction as the general low-latitude pattern, driven by convective mixing [7]. Dust devils were detected repeatedly from mid-morning through late afternoon, with marked temperature and pressure signatures as they passed the lander. Those figures sit inside the ranges later compiled across all landed sites, where measured air temperature spans 140 to 300 K and background column dust optical depth stays below 1 outside storm season [7].

Background column dust optical depth over the aphelion half of the Martian year sits below 1, and the opacity of 0.5 measured through these 83 sols is within that range rather than elevated by regional activity [7].

Ground operations

The mission returned 83 percent of a projected objective across 230 commanded maneuvers [4]. The command cycle was one uplinked plan per sol executed against onboard hazard handling, because the one-way light time exceeded ten minutes; this is the same cycle that limited Go To Waypoint commands to running without oversight for their full length, described above [3]. Eleven Wheel Abrasion Experiments were run inside that cycle, each costing seven wheel revolutions and a stationary interval with five wheels locked, which is why the total across the mission is only 39 revolutions of instrumented wear [4].

The WAE strips carried five film thicknesses of each metal on one substrate, so a single run returns a wear-versus-thickness curve rather than a single wear datum, and the inverse relationship observed for aluminum is what identifies abrasion rather than contamination as the cause of the reflectance change.

Technologies developed

Rocker-bogie mobility, onboard hazard avoidance from stereo plus laser ranging, and the sol-by-sol command cycle were all first flown here and are carried by every later Mars surface mission [4], [8]. The dig-in depths it measured, 0.6 to 1.5 cm depending on material, are the earliest in-situ constraint on the bearing behavior later quantified as cohesion at or below 1 kPa for sand and up to 11 kPa for indurated soils [10]. The autonomy split it recorded, weak dead reckoning against workable hazard avoidance, set the target the next generation of Mars rovers designed against: roughly 100 m of autonomous traverse per sol, about what Sojourner covered across its entire mission [3].

The Wheel Abrasion Experiment established a method that has not been repeated: using the rover’s own wheel as a tribometer, with a graded set of metal films as the calibrated reference, to bracket the hardness of the surface material. It cost 10 g of detector and one modified wheel [12]. Its accidental detection of electrostatic charging, and the grain size limit that follows from it, is a result no imaging or spectroscopic instrument on the vehicle could have produced [9].

What is not established

The soil-mechanics cohesion figures derived from wheel-spin experiments rest on an assumption, that the friction angle equals a photographically measured angle of repose, that the same analysis’s own least-squares fits do not consistently satisfy, so the reported cohesions of a few tenths of a kilopascal should be read as illustrative rather than as a tight bound; bulk density for the site was never measured directly, but inferred from friction angle through a correlation built for lunar soils and simulants and applied here without an error estimate [9]. The conglomerate origin proposed for the rounded surface objects is one of several competing explanations offered from imagery alone, with no compositional or textural measurement to distinguish among them [9]. The operations figures for traverse rate and dead reckoning error come from a single mission on one vehicle at one site, reported by the mission’s own engineers rather than measured against an independent baseline [3].