The Design and Engineering of Curiosity

by Emily Lakdawalla

  3.5.1 Site context

  Curiosity landed in the northern floor of Gale crater, at 4.5895°S, 137.4417°E and an elevation of 4501 meters below the Martian datum (Figure 3.3). One of the deepest holes on Mars, Gale is located at the boundary between Mars’s southern highlands and northern lowlands. Gale displays clear evidence for water having once flowed from the highlands surrounding the crater through gaps in the rim and then depositing overlapping alluvial fans of sediment on the crater floor. One such channel and fan is Peace Vallis, to the northwest of the landing site. But there were many other such channels and fans all around the rim; Peace Vallis was just one of the last to form, so is the most prominent today.

  Figure 3.3. Context map of the landing site. Inset: topographic map of Gale crater from Mars Express. Gale is about 150 kilometers in diameter. Yellow rectangle shows location of main map. Elevation scale is in meters relative to Martian datum. Main map: major landmarks at the landing site. Base image is the Mars Odyssey THEMIS daytime infrared mosaic. NASA/JPL-Caltech/ASU/Emily Lakdawalla.

  At the center of Gale crater is a 5-kilometer-tall central mound of layered sediments formally named Aeolis Mons. The science team refers to the mountain as Mount Sharp, after Robert Sharp, a pioneering Caltech planetary geologist. Researchers studying orbital data before the landing were divided on how the mountain formed, but it did seem clear that different styles of geology prevailed at different times. In particular, the lowermost elevations of the mountain were made of nearly horizontally layered rocks, whereas the upper, brighter slopes lacked such obvious layering. NASA’s Mars Reconnaissance Orbiter had spotted spectral signs of clays, sulfates, and hematite in Gale’s lowermost layered rocks, all of which form in different kinds of wet environments. Reaching the lowermost slopes of the mountain to study those rocks was a major goal for the science team. Then they hoped to climb up through the layered rocks to study the history preserved in them. They never intended or expected to summit the mountain; at best, they hoped to reach the boundary between the lower mound and the upper mound in order to understand what happened in Mars’ history to change the style of sediment deposition so abruptly.

  During the cruise, members of the science team had carefully mapped the geology of the landing ellipse using orbital images, so they were prepared with a good understanding of the regional geology and their likely drive route before the landing. The landing placed them several kilometers to the southeast of the end of the Peace Vallis fan. Between the rover and the mountain lay a linear swath of black sand dunes, later called the Bagnold dune field, named for Ralph Alger Bagnold, a pioneer in the study of sand’s behavior in deserts. Most of the dune field was considered hazardous to the rover. However, southwest of the landing site, the dunes thinned out near a cluster of steep-sided buttes that the mission named the Murray buttes after Bruce Murray, a Mars geologist and early leader in NASA’s Mars exploration. To reach the interesting rocks at the base of the mountain, Curiosity faced a lengthy drive – more than 9 kilometers as the orbiter flies, much longer for a wheeled rover dodging obstacles. At Murray buttes, the rover could cross the dune field and finally reach the lower mound’s layered rocks. Figure 3.4 is a visual summary of the rover’s route.

  Figure 3.4. Route map for the mission to sol 1800. Bold text denotes drill or scoop sites. Map by Emily Lakdawalla on a base image of Mars Reconnaissance Orbiter CTX mosaic colorized with Mars Express HRSC image.

  3.5.2 Yellowknife Bay campaign and the sol 200 anomaly

  The first order of business upon landing was to establish the basic functions of the rover, like raising the mast, establishing routine telecommunications, and making sure that the rover’s power and thermal systems were operating as expected. Then the rover stood down for four sols for an upgrade of its operating system, reprogramming the main computer from an interplanetary spacecraft into a surface rover. Figure 3.5 provides an overview of the major external Curiosity rover systems relevant to the surface mission.

  Figure 3.5. Science instruments and major external systems of the Curiosity rover. Top image is a rover self-portrait taken at John Klein on sol 177. NASA/JPL-Caltech/MSSS image release PIA16764. Bottom image taken June 3, 2011 during mobility testing. NASA/JPL-Caltech image release PIA14254, annotated by Emily Lakdawalla.

  Initially, the engineering side of the tactical team had more control over Curiosity than the science team did. Curiosity required a commissioning activity phase to work it through its engineering and science functions. Even after commissioning ended, there was a long list of first-time activities that engineers methodically paced through: first drive, first contact science target, first scooping, first use of different driving modes, first drill site, and so on.

  Rather than immediately beginning the journey southwest across nondescript-looking terrain, the project science group decided to start the mission with a drive of about 500 meters east to a location they named Yellowknife Bay, where three distinct rock types occurred together. While working through its first-time activities, Curiosity would be able to perform productive science there.

  Curiosity used all the environmental and remote sensing instruments for the first time on Mars at the landing site. RAD measured the radiation environment. DAN detected neutrons from the ground (and also from the MMRTG). Mastcam took photos, testing out its focal mechanism. ChemCam lasered a rock. REMS took weather data. In the very first days of operations, the REMS team discovered that one of their wind sensor booms had been damaged, likely by gravel launched into the air by the force of the descent engines impinging upon the surface. The wind experiment on REMS has never functioned fully, but the rest of the REMS sensors have been active since landing.

  The rover drove for the first time on sol 16. During the next few weeks the rover drivers slowly increased its driving autonomy, beginning with blind drives, later adding in visual odometry to increase drive accuracy (see section 6.​5 for more on the different driving modes). SAM ingested its first atmospheric sample on sol 18. The SAM team initially thought they had detected abundant methane in the air, but it turned out to be contamination from the leak that had happened before launch (see section 9.​5.​1.​3).

  Scientists selected a rock they named Jake Matijevic for the first contact science observations (APXS compositional info and MAHLI photos) on sol 46. The engineers commanded the arm to reach out and scoop a sample of sand for the first time at the Rocknest sand drift on sol 61 (Figure 3.6). They shook the sample inside the Collection and Handling for In situ Martian Rock Analysis (CHIMRA) sample handling mechanism on the arm in order to scrub the apparatus of any remaining manufacturing residue.8 They scooped again and delivered the first Martian samples to the laboratory instruments CheMin and SAM for the first times on sols 71 and 93.

  Figure 3.6. Left Mastcam photo 0061ML0003060000102375E01 documenting the first day of scooping at Rocknest. CHIMRA acquired a full scoop from the site at lower right, then shook out some of the sample to reduce the amount in the scoop, leaving a fresh pile of dark sand on the ripple surface. NASA/JPL-Caltech/MSSS.

  MAHLI captured the first full rover self-portraits at Rocknest on sols 84 and 85 (see section 7.​4.​3.​5 for more on how MAHLI captures self-portraits). In order to continue analyzing the Rocknest sample while proceeding toward Yellowknife Bay, the rover planners developed and quickly deployed the ability to drive with cached sample held inside the CHIMRA mechanism (see section 5.​4.​5).

  Curiosity acquired its first drilled sample on sol 176, at John Klein (Figure 3.7). Analyses of the first drill samples were interrupted by a major anomaly – arguably the scariest event of the mission after landing – on Wednesday, February 27, sol 200. The event is now known as “the sol 200 anomaly.” The routine morning uplink revealed that the rover was behaving strangely, returning real-time telemetry but not performing commanded activities. Engineers quickly diagnosed an issue with the rover’s onboard memory. Later in the day, their concern elevated when more telemetry from Curiosity indicated that it h
ad not gone to sleep as commanded, so was depleting its batteries.

  Figure 3.7. John Klein drill site after drilling activities were completed. Mastcam acquired this photo on sol 229, after CHIMRA dumped the remaining drilled sample in two piles. NASA/JPL-Caltech/MSSS release PIA16815.

  Unlike the smaller Spirit and Opportunity rovers, Curiosity has an entire spare computer system available for the rover to switch to. On sol 201, the mission uplinked commands to swap from the A-side computer, with its corrupted memory, to the B-side computer. Commissioning of the instruments on the B-side computer was completed as of sol 223 (23 March 2013). The rover has operated on the B-side computer ever since.

  As a consequence of the swap to the B-side computer, the rover switched eyes. It has two pairs of each of its engineering cameras (the Navcams and front and rear Hazcams, see section 6.​3), with complete sets connected to each computer. The switch from A-side to B-side computers moved its Navcam point of view down by 4.8 centimeters. Similarly, the front Hazcam view of the world shifted to the rover’s left by 8.2 centimeters. The rear Hazcam view shifted to the rover’s left by about a meter, from one side to the other of the MMRTG. Because the rover’s autonomous hazard-finding software had been trained on Mars only with A-side images, the project was forced to repeat some of the commissioning activities using the rover’s new eyes.

  During recommissioning of autonomous driving modes, the engineers made the unpleasant discovery that the pointing of the B-side Navcams shifted very slightly over the course of a sol, likely because their mounting bracket warped with daily extremes of Martian temperature, enough to confuse the onboard rangefinding software. Over the ensuing weeks, they had to perform calibration activities to understand the temperature-dependent behavior. Only after this investigation was complete were they able to test autonomous navigation capability.

  After recovering from the sol 200 anomaly, rover operations almost immediately stood down again because of solar conjunction. When the Sun is within 3° of Mars in Earth’s sky, radio communications can be affected by solar radio emissions. Mars landers and orbiters aren’t directly affected, but because communications aren’t reliable, Earth controllers avoid any activities that might place the spacecraft at risk of needing intervention. During solar conjunction, from sol 235 to 260, the rover performed only background environmental science observations and transmitted a daily “beep” to Earth. After conjunction, the rover drilled for a second time at a nearby site named Cumberland, on sol 279.

  Early impressions of the drilled material suggested that Curiosity had accomplished its science objectives (listed in Box 1.​5). The mission had successfully explored the biological potential of at least one target environment (using SAM to inventory organic compounds) and had gathered the data needed to conclude that the environment was a biologically relevant one (the still water of a lake bottom). The mission had characterized the regional geology of the landing site before landing, and followed that up with successful chemical, mineralogic, and isotopic analyses with the science instruments. The isotopic measurements of water in the ancient Mars rocks had corroborated orbital science results indicating that Mars has lost much of its atmosphere. And RAD’s successful operation had hit Curiosity’s last goal of characterizing surface radiation. With all the crucial first-time activities complete and minimum mission success achieved, the science team could go on their driving adventure.

  3.5.3 The Bradbury traverse

  On sol 295, Curiosity departed Cumberland, investigating a few outcrops close to Yellowknife Bay. Then the rover embarked on a 13-kilometer journey across the floor of the crater to the southwest, toward Murray buttes and the gap in the Bagnold sand dune field that would allow the rover to cross it safely and reach the base of the mountain.

  During the time at Yellowknife Bay, engineer Paolo Bellutta had led the effort to map out a “rapid transit route” for the rover using the traversability algorithms he had developed during the landing site selection process (see section 1.​6.​3). It was not the shortest possible path, but sought to minimize drive time by keeping the rover to relatively flat terrain with good visibility, which would maximize single-sol drive distances. (Being able to see far ahead permits longer blind drives, which is the fastest driving mode; see section 6.​5 for more about the different driving modes.) The traverse was across a region of low, hummocky plains with rare outcrops of rock. From orbit it appeared largely similar to the terrain Curiosity had already traversed. The rover would be permitted to perform science observations as opportunities came up, but driving was a higher priority than science. They used autonav for the first time to extend a planned drive on sol 347, and quickly racked up record-breaking drive distances, including one of 141 meters on sol 385.Long drives also used up time and energy, limiting resources available for science.

  The science team selected three locations along the proposed path where orbital images suggested that there was more coherent outcrop, worthy of brief stopovers for science. Curiosity reached the first site, Darwin, on sol 390, staying until sol 402. On sol 426, the rover passed from terrain that the science team had mapped as “hummocky plains” to a new landscape, called “rugged terrain” (Figure 3.8). Rugged terrain featured more bedrock in the form of sharp blocks of rock protruding from the plains. The rougher terrain slowed down autonav, making drive distances shorter. On sol 439, Curiosity approached a site called Cooperstown to characterize the rugged-terrain rock.

  Figure 3.8. Top: typical hummocky terrain. Part of a mosaic of left Mastcam images from sol 412. Bottom: typical rugged terrain. Part of a mosaic of left Mastcam images from sol 437. NASA/JPL-Caltech/MSSS.

  The months after Cooperstown were full of problems. An unsuccessful flight software update (see section 4.​3.​2.​2) delayed them at Cooperstown until sol 453. On sol 456, the rover experienced a “soft short” in its MMRTG, later determined to have been caused by part of the electrical power circuit touching its metal housing (see section 4.​2.​3).9 The short spontaneously resolved itself on sol 461 and didn’t recur for a year. On sol 463, the rover drivers commanded a set of MAHLI images of the wheels, which revealed a huge hole in the left front wheel. They started commanding wheel images after every drive to monitor the development of the wheel damage. Wheel imaging slowed down driving and revealed rapidly progressing damage (Figure 3.9 and section 4.​6.​4).

  Figure 3.9. Development of damage to the left front wheel. MAHLI images taken on sols 177, 411, 463, and 469. NASA/JPL-Caltech/MSSS.

  The mission appointed a Tiger Team led by Rich Rainen (who had managed the rover’s mechanical team during its construction) to answer three questions: What was causing the damage? How could the mission reduce or prevent further damage? And what was the life expectancy of the wheels?10 Following experiments in the Mars Yard, the Tiger Team quickly determined that the rugged terrain was a factor. Sharp-pointed rocks that were embedded in the ground did not shift when the rover passed over them; instead, they pierced the wheels. No rover had encountered such embedded, sharp rocks before.

  The project directed the engineers to avoid pointy rocks to the best of their ability, take one set of wheel images on every drive, and perform full wheel imaging (five sets of images interspersed with 60-centimeter drives, in order to present all surfaces of the wheels to the cameras) once every 100 meters. This effectively ended the use of autonav for some time, which dramatically slowed the rover’s progress. Moreover, every full-wheel-imaging sol advanced the rover only 2 meters toward the destination at the cost of a precious drive sol. One consolation was that sequencing arm activities for wheel imaging permitted more opportunities for APXS and MAHLI use than the science team had previously been able to justify. The project began to employ surge sols (see section 3.4.3) in order to make the most of every opportunity to drive.

  As the engineers performed further tests on wheel damage, a group of scientists and engineers led by John Grotzinger and Matt Heverly mapped the terrain ahead and drove a virtual rover thro
ugh digital terrain models to develop plans for a feasible future long-term drive path that would avoid wheel-damaging terrain. Unfortunately, the previously planned rapid transit route, which had preferred high ground for visibility, coincided with the worst terrain. In between high ground were depressions filled with sand, which would be kinder to the wheels, but the valleys had their own problems: driving in depressions meant less long-distance visibility; required a slightly longer drive distance in order to detour around highlands; and had potential issues with “pinch points” where the rover would have to pass through relatively narrow and/or steep gaps in order to exit one valley and enter another.

  On sol 524, the rover departed the rapid transit route to enter sand-filled valleys. The rover had to pass over a relatively high sand ripple blocking Dingo Gap in order to enter the first of the valleys (Figure 3.10). It successfully made the crossing on sol 535.

  Figure 3.10. Dingo Gap, where a tall sand ripple obstructed Curiosity’s progress westward into the safer valleys. In the distance is the rim of Gale crater. Left Mastcam mosaic from sol 530. NASA/JPL-Caltech/MSSS.

  Driving in valleys provided far more opportunities to study rock layers from the side, and the science observations improved. On sol 574, the rover approached the third and final Bradbury traverse science stop, named the Kimberley, where three distinct rock units came together (Figure 3.11). They spent nearly two months at the Kimberley, drilling at Windjana on sol 621. While working at Windjana, the MAHLI instrument experienced and recovered from its first anomaly, which put the instrument out of service from sols 615 through 626.