In This Issue
Fall Issue of The Bridge on Space Exploration
September 1, 2021 Volume 51 Issue 3
Close collaboration between engineering and science has enabled marvels of space exploration over decades. Eight exemplary missions are described in this issue, conveying the excitement, challenges, and breakthroughs involved in efforts to better understand the wonders and mysteries of this solar system.

Collecting a Sample from Asteroid Bennu: Science and Engineering Enable OSIRIS-REx

Tuesday, September 14, 2021

Author: Heather L. Enos and Michael C. Moreau

The diverse science and engineering perspectives and experiences of the OSIRIS-REx team were key to overcoming the challenges of asteroid Bennu.

On October 20, 2020, the OSIRIS-REx  spacecraft began a maneuver that had been envisioned years prior by scientists and engineers at the University of Arizona, NASA’s Goddard Space Flight Center, Lockheed ­Martin, and KinetX Aerospace (Lauretta et al. 2017, 2021). For the preceding ­several weeks, the 1400 kg (3000 lb) spacecraft had slowly orbited near-Earth ­asteroid Bennu from 600 meters (0.37 mile) above the surface, with ground controllers at Lockheed Martin’s facility in Littleton, Colorado, carefully tweaking the plane and phasing of the orbit with millimeter-per-second precision to set up for this moment.

The mood in the OSIRIS-REx Mission Support Area at Lockheed Martin was tense, but with excitement—just a few hours from an event that had been more than 14 years in the making: the first attempt by NASA to collect a bulk sample of material from the surface of an asteroid for return to Earth.

A few hours earlier, command sequences had been loaded on board the spacecraft with final adjustments to the critical parameters that would guide the spacecraft to contact with the surface of tiny Bennu (its mean diameter is less than 500 m). From this point on, members of the operations team in Denver could only listen for bits of telemetry sent back as the spacecraft autonomously navigated through the sample collection sequence.

Sampling on the Asteroid

At about 10:30 am Mountain Standard Time, the spacecraft fired its attitude control thrusters to perform a maneuver resulting in a 6 cm/s velocity change. In the microgravity environment of Bennu, this small velocity change caused the spacecraft to make a hard right turn, leave the safety of the terminator plane orbit, and venture out over the sunlit side of Bennu.

Enos and Moreau fig 1.gif
FIGURE 1 OSIRIS-REx mission timeline. The spacecraft spent over 2 years in close proximity to Bennu, first conducting detailed surveys and reconnaissance passes to characterize the asteroid and select the TAG site, then preparing for the sample collection activity. It departed Bennu in May 2021 and will return the sample capsule to Earth in September 2023. TAGSAM = Touch and Go Sample Acquisition Mechanism. Image credit: NASA/Goddard/University of Arizona.

A 3-meter-long robotic arm was deployed, with a mechanism for sample acquisition attached to the end (figure 1). The Touch and Go Sample Acquisition Mechanism (TAGSAM; Bierhaus et al. 2018) was specifically designed to solve the unique problem of ­sample acquisition from the microgravity environment of a small asteroid. It acts as a reverse vacuum cleaner, with high-pressure nitrogen gas mobilizing surface material that is then forced into the containment area of the “head.” “Touch and go” describes the concept of operations for sample collection: the TAGSAM head is in contact with the surface for only a few seconds.

The spacecraft next pointed its navigation camera toward Bennu and began recording images to process in the natural feature tracking autonomous navigation system. About 4 hours after leaving orbit, the spacecraft slewed to the sample collection attitude with the TAGSAM arm and cameras pointed toward the surface. At about 125 meters above the surface, the spacecraft fired its thrusters again to start a steep descent to the surface at a velocity of about 16 cm/s. Ten minutes later and at an altitude of about 40 meters, a 6 cm/s maneuver slowed the spacecraft to a descent rate of 10 cm/s and matched the spacecraft velocity with the surface of the rotating asteroid below.

While these final maneuvers were being computed, the predicted spacecraft trajectory was propagated forward to the time of surface contact and compared against a map of the hazardous locations on the surface. As the spacecraft passed below 5 meters altitude, it performed a final check against the hazard map—and the predicted contact point was safe! Seconds later the TAGSAM touched down on Bennu (figure 2) and began to penetrate the surface. The spacecraft continued its downward velocity as a storm of particles and dust were mobilized around and inside the TAGSAM head. Just 6 seconds after contact was detected the spacecraft fired its back-away thrusters to arrest its descent and leave the surface.

Enos and Moreau fig 2.gif
FIGURE 2 Contact with Bennu: Two images from the OSIRIS-REx SamCam imager, taken about 2 seconds apart. The frame on the left was shuttered just after initial contact with the surface of Bennu and milliseconds before the TAGSAM high-pressure nitrogen gas bottle fired; the image on the right shows the scene just milliseconds after the TAGSAM gas bottle fired. Image credit: NASA/Goddard/University of Arizona.

By all accounts the sample collection event executed flawlessly, likely yielding hundreds of grams of ­asteroid material captured in the TAGSAM. The sampling event and subsequent firing of thrusters to back away excavated an area covering several square meters to a depth of half a meter or more, and dislodged meter-scale boulders in the vicinity of the contact point. The spacecraft contacted less than 1 meter from the targeted location, avoided all nearby hazards, and touched the surface in a place that had an abundance of fine-grained material ingestible by the TAGSAM. The TAGSAM sank into the regolith several centimeters to ensure that asteroid particles would be forced into the head once the high-pressure nitrogen gas bottle fired. The surface of the asteroid responded in a ­dramatic fashion to the imparted forces, and scientists will be studying the resulting imagery for years to come to gain insights into the regolith properties and cohesion of the surface.

Importance of Science and Engineering Partnership

Notwithstanding thorough preparation, the path to successful sample collection at Bennu was littered with surprises and obstacles. Since the earliest days of the mission, the OSIRIS-REx team has benefited from the tight integration of science and engineering disciplines to inform flight system and mission design, interpret surprises, and respond to the unexpected.

It is not uncommon on a NASA mission for scientists and engineers to work closely together. The scientists typically pose the key questions about the natural world or the universe that the mission seeks to answer, and draw conclusions from the data collected; the engineers develop the machinery and systems to accomplish the observations with an eye toward optimizing performance and minimizing risk.

A first example of this partnership between science and engineering began before OSIRIS-REx was selected as a NASA mission. Extensive Earth-based observations of Bennu, combined with scientific models of the asteroid, were used systematically to constrain over 100 asteroid parameters, covering orbital, bulk, rotational, radar, photometric, spectroscopic, thermal, regolith, and environmental properties. These scientific data were essential to accurately plan the encounter with the asteroid and specify the key performance parameters of the spacecraft before its assembly and launch, and helped ensure that the spacecraft had sufficient capabilities to handle some of the surprises encountered.

We examine some of the best examples of science and engineering disciplines working together to enable the success of OSIRIS-REx.

Adapting to Mission Challenges and Surprises

The unique nature of a sample return mission to a very small planetary body created significant operational challenges. These included accurate spacecraft navigation in the microgravity environment, precision delivery of the spacecraft to the asteroid surface, and schedule-driven scientific data generation based on operational milestones to enable sample-site selection. Good communication and coordination between science and engineering team members were essential.

The small size of Bennu and corresponding large navigational uncertainties meant that science planning and implementation tools required very specialized capabilities. As multiple off-the-shelf planning tools were investigated, the team came to the conclusion that a customized tool was required. The J-Asteroid tool (Christensen et al. 2018) implemented capabilities and features unique to each of the observational phases at Bennu.

Moreover, the overall planning of science observations required tight coordination between navigation team members and science planners to balance operational complexity with science return. Navigation team members needed to understand the objectives of the science motivating particular observations. ­Science planning team members needed to understand the navigational uncertainties and how to trade between improving science value and minimizing the risk of data gaps. Successful science observation planning required a sophisticated understanding of these concepts, and team members had to spend a lot of time outside their traditional lanes.

Another complicating factor was that most science observations during proximity operations at Bennu required a “late update” to compensate for navigation uncertainties. A late update involved a complex handoff of the products required to generate updated spacecraft command sequences between the navigation team, science operations team, and spacecraft control center in the 24 hours leading up to execution of every maneuver or science observation to retarget the precise pointing and timing.

The TAGSAM sample collection mechanism and the performance of the whole flight system to target a TAG location on the surface were based on a fundamental assumption that Bennu was similar to a sandy beach, uniformly covered in relatively fine-grained material, and with large flat areas suitable for safely navigating the spacecraft to the surface. This understanding was developed largely from an extensive Earth-based observational campaign conducted before launch: low observed thermal inertia pointed to a uniform surface and fine-grained particles (Emery et al. 2014). TAGSAM was designed to ingest particles 2 cm and smaller (Bierhaus et al. 2018).

Enos fig 3.gif

The flight system was designed to set down within a 50-meter diameter landing zone. But imagery recorded by the PolyCam (Rizk et al. 2018) during the spacecraft’s approach in November 2018 was a shock to the team: Bennu was littered with boulders, from meter-scale to some the size of multistory buildings (figure 3). In fact, in the first global mosaic of Bennu it wasn’t clear that there was sampleable material anywhere on the asteroid. There were certainly no hazard-free locations that came even close to the 50-meter diameter landing zone the spacecraft had been designed to target. The correlation between low thermal inertia and fine-grain surface material was false, for reasons that may not be fully understood until the Bennu sample is analyzed on Earth.

The realization of an extremely rocky Bennu dramatically affected the plans for the site selection and TAG. To land safely in a location with sampleable material, the spacecraft would have to make contact in an area 10 times smaller than the system was designed for—essentially TAG in the center of a bull’s-eye—requiring a new onboard navigation method and a major effort to fine-tune the performance of the spacecraft.

This was just one of many curveballs thrown by asteroid Bennu! The next was within days of the first orbit insertion on December 31, 2018, when navigation images showed hundreds of millimeter- to centimeter-scale particles emanating from the asteroid’s surface (Lauretta et al. 2019), potentially creating a hazard for the spacecraft. The activity is modest enough that it could only have been detected by a nearby spacecraft (not by telescopes). It was a huge surprise and an exciting scientific discovery.

Safety was the highest priority, even if it meant there was a possibility of collecting less sample.

Analysis revealed that some particles escape into space, others temporarily orbit the asteroid, and most fall back onto Bennu’s surface (Hergenrother et al. 2020). The management team scrambled to verify that it was safe for the spacecraft to remain in orbit. The science and navigation teams developed independent tools and techniques to attempt to characterize the size and origin points of particle ejections. Members of the navigation team developed autonomous techniques to identify and track these particles, while navigation and science team members collaborated to fit orbits to individual particles (Hergenrother et al. 2020). With a better understanding of the particle ejections, it was determined that the spacecraft was safe.

The serendipitous discovery of Bennu’s activity would not have happened if not for the particular application of optical navigation adopted by the team and the fact that the spacecraft was just a couple of kilometers from the surface. The navigation concept called for the use of a very wide field-of-view (FOV) camera, 33 × 44 degrees—comparable to a Go-Pro—and used short and long exposures to resolve both Bennu (short) and stars (long) in pairs of immediately successive images. The design decision to continue the use of long-exposure stellar imaging for pointing corrections after the spacecraft entered orbit and the selection of the wide FOV camera both enabled the particle detections.

The particles contributed to another scientific bonanza for the team: their reconstructed orbits made it possible to estimate the gravity field of Bennu—a key science objective—to a fidelity that could not have been achieved with observations of the spacecraft’s orbit alone (Scheeres et al. 2020). The discovery of the particle ejection phenomenon and the ability to recognize and interpret the scientific implications would not have been possible without the tight integration between science and engineering on the mission—and this turned out to be a trial run for the type of collaborative effort required for the site selection and TAG.

Scientific Observation and Criteria for Sample Site Selection

To select a sample site on Bennu’s rocky terrain, coordi­nated observations were combined into four thematic maps of decision-making properties: deliverability, ­safety, sampleability, and science value. The site selection team considered the probability of success as indicated by each map; the best site would be where the maps intersected, indicating the best chance of collecting a sample on the first TAG attempt.

The much more rugged surface than predicted created a significant challenge in identifying potential sample sites that satisfied the deliverability and safety criteria. Only a small number of hazard-free regions, about 5–8 m in radius, were identified.

As the site selection team assessed the candidate TAG sites, the engineering team worked to enhance the capabilities of the spacecraft, allowing more of the potential sites to become viable options for sample collection. Refined deliverability and safety maps led to revised relative scoring of the candidate sites. Sample sites that had the highest probability for safely contacting the surface rose to the top. Safety was the highest priority, even if it meant there was a possibility of collecting less sample.

The team then focused on the sampleability of the potential sites. The objective was to select a site that offered a high probability of collecting at least 60 grams of material on the first attempt. As described earlier, Bennu’s uniform thermal inertia meant that such data could not be a discriminator to identify centimeter-scale particles.

Finally, careful mapping of resolved and uningestible particles provided the quantitative assessment of sampleability of four candidate sites (Burke et al. 2021; Lauretta et al. 2021), two of which, known as ­Nightingale and Osprey, were selected as the primary and backup sampling sites, respectively.

Engineering the TAG at Nightingale Site

The Nightingale sample site was only about 16 meters across, much less than the 50-meter diameter TAG zone for which the spacecraft had been designed (figure 4). The TAG design had originally incorporated a simple lidar-based range update to correct trajectory errors during the descent to the site. Prelaunch analysis showed that lidar-based navigation would be sufficiently accurate for a 50-meter diameter landing zone, but not for the much smaller area.

Enos and Moreau fig 4.gif
FIGURE 4 Visualization showing the size of the Nightingale TAG zone (blue) compared to the original area the spacecraft was designed to target (orange), with a sketch overlaid of typical parking lot spaces to provide scale. A large rock formation at about 4 o’clock in the image is the height of a two-story building. Image credit: NASA/Goddard/University of Arizona.

To execute a bull’s-eye TAG, the team pursued three parallel activities. First, the lidar-based technique was abandoned in favor of natural feature tracking (NFT), originally developed as a backup to the lidar. The NFT capability relied on optical navigation images and preplanned maps for landmark navigation (MLNs).

For the original mission sufficient imagery was collected to generate and validate maps for NFT use at orbital altitudes (600–1000 m), but the use of NFT during TAG would require much higher resolution maps that could be used all the way to the surface. While NFT would provide more precise navigation updates, it required a whole new effort to conduct additional high-precision observations of the candidate TAG sites and develop high-resolution MLNs. This effort involved contributions from the navigation and science planning teams to design new close reconnaissance observations of the primary and backup TAG sites, and from the spacecraft guidance, navigation, and control team and science altimetry working group to develop and validate the MLNs for use during TAG.

The second activity involved work by the navigation and spacecraft teams to revise and refine all of the analysis and modeling of the spacecraft to extract maximum possible performance. Thanks to extensive characterization of Bennu’s gravity, solar radiation pressure, thermal reradiation forces, and other small forces acting on the spacecraft, the navigation team improved the predictive accuracy from tens of meters to only a few meters. Navigation and spacecraft team members calibrated the performance of the spacecraft thrusters such that velocity change maneuvers were performed with millimeter-per-second precision.

These modeling improvements pushed the envelope for deep space navigation performance and reduced the levels of error for TAG to be very close to the bull’s-eye goal. But the team still had to deal with the problem that any small error causing the spacecraft to drift off course could be potentially disastrous given the hazardous rocks surrounding the targeted TAG location.

The third activity therefore related to protecting the spacecraft from hazards. Centimeter-level precision topographic maps developed by the science altimetry working group made it possible to map all the rocks and slopes that could damage the spacecraft during TAG. The hazard map can be thought of as a grid of points overlaid on the topographic map of the site; grid points that represented a hazard were shaded red and those that were safe green. This hazard map was implemented on the spacecraft as a final check during TAG (figure 5): the spacecraft could proceed to the surface only if it was projected to contact a green area.

Enos and Moreau fig 5.gif
FIGURE 5 A successful “touch and go” (TAG) at Nightingale required both an autonomous onboard navigation system, known as Natural Feature Tracking (NFT), and an abort capability to keep the spacecraft safe if it predicted contact in a hazardous location. The left portion of the graphic illustrates the concept for NFT, in which preidentified topographic features are located in optical navigation images and used to update the spacecraft position and velocity. The right portion of the graphic shows two scenarios involving the hazard map: if the NFT system predicts contact in an area of the map shaded green it will proceed to contact and sample collection; if contact is predicted in a red area the result will be an abort 5 meters above the surface. Image credit: University of Arizona.

The entire site selection process entailed dedicated teamwork and a carefully calibrated interplay between the capabilities of the spacecraft to touch down in a hazard-free location and the characteristics of the selected TAG site.


When OSIRIS-REx successfully collected its asteroid sample, it was the culmination of extensive collaboration between scientists and engineers working to overcome the challenges presented by tiny asteroid Bennu—those that were anticipated and those that were a surprise to the team.

Any successful mission of planetary exploration involves a very high degree of teamwork and complicated interfaces between the scientists focused on collecting certain observational data, and the engineers responsible for designing and operating the spacecraft and keeping it safe. But it is uncommon for these disciplines to work as closely together as was the case on OSIRIS-REx, and the diversity of perspectives and experiences of this broader team were key to over­coming the challenges of Bennu.

On May 10, 2021, OSIRIS-REx departed Bennu, commencing a 2-year, 289 million km journey to return its precious asteroid sample cargo to Earth in 2023. Analysis of the sample will advance understanding of these primitive bodies and the very origins of this solar system.


OSIRIS-REx spacecraft operations are performed by Lockheed Martin, in Littleton, Colorado. The navigation team consists of engineers from KinetX Aerospace (Simi Valley, CA) and Goddard Space Flight Center (Greenbelt, MD). The University of Arizona in ­Tucson hosts the science planning, science operations, and science data process facilities. Goddard Space Flight ­Center provides mission management.


Bierhaus EB, Clark BC, Harris JW, Payne KS, Dubisher RD, Wurts DW, Hund RA, Kuhns RM, Linn TM, Wood JL, and 4 others. 2018. The OSIRIS-REx spacecraft and the Touch-and-Go Sample Acquisition Mechanism (TAGSAM). Space Science Reviews 214(7):107.

Burke KN, DellaGiustina DN, Bennett CA, Walsh KJ, Pajola M, Bierhaus EB, Nolan MC, Boynton WV, Brodbek JI, Connolly HC, and 19 others. 2021. Particle size-frequency distributions of the OSIRIS-REx candidate sample sites on asteroid (101955) Bennu. Remote Sensing 13(7):1315.

Christensen PR, Anwar S, Burris ME, Carter SR, ­Dickenshied SL, Noss DD, Hagee WR, Rios KJ, Wren PF. 2018. J-­Asteroid: JMARS for asteroids and other small bodies. LPI Contrib. No. 2083, 49th Lunar and Planetary Science Conf, Mar 19–23, The Woodlands TX. 

Emery JP,  Fernández YR, Kelley MSP, Warden (née Crane) KT, Hergenrother C, Lauretta DS, Drake MJ, Campins H, Ziffer J. 2014. Thermal infrared observations and thermophysical characterization of OSIRIS-REx target asteroid (101955) Bennu. Icarus 234:17–35.

Hergenrother CW, Adam CD, Chesley SR, Lauretta DS. 2020. Exploration of the activity of asteroid (101955) Bennu. Journal of Geophysical Research: Planets 125(9):e2020JE006549.

Lauretta DS, SS Balram-Knutson SS, Beshore E, Boynton WV, Drouet d’Aubigny C, DellaGiustina DN, Enos HL, Golish DR, Hergenrother CW, Howell ES, and 38 others. 2017. OSIRIS-REx: Sample return from asteroid (101955) Bennu. Space Science Reviews 212:925–84.

Lauretta DS, Hergenrother CW, Chesley SR, Leonard JM, Pelgrift JY, Adam CD, Al Asad M, Antreasian PG, Ballouz R-L, Becker KJ, and 47 others. 2019. Episodes of particle ejection from the active asteroid (101955) Bennu. Science 366(6470):eaay3544.

Lauretta DS, Enos HL, Polit AT, Roper HL, Wolner CWV. 2021. OSIRIS-REx at Bennu: Overcoming challenges to collect a sample of the early solar system. In: Sample Return Missions, ed Longobardo A. Amsterdam: Elsevier.

Rizk B, Drouet d’Aubigny C, Golish D, Fellows C, Merrill C, Smith P, Walker MS, Hendershot JE, Hancock J, ­Bailey SH, and 45 others. 2018. OCAMS: The OSIRIS-REx ­camera suite. Space Science Reviews 214:26.

Scheeres DJ, French AS, Tricarico P, Chesley SR, Takahashi Y, Farnocchia D, McMahon JW, Brack DN, Davis AB, ­Ballouz R-L, and 38 others. 2020. Heterogeneous mass distribution of the rubble-pile asteroid (101955) Bennu. ­Science Advances 6(41):eabc3350.


About the Author:Heather Enos is OSIRIS-REx deputy principal investigator, University of Arizona. Michael Moreau is OSIRIS-REx deputy project manager, NASA Goddard Space Flight Center.