Keyhole-Based Site Selection for Kinetic Impact Deflection of Near-Earth Asteroids

Given present-day asteroid discovery capabilities, near-Earth asteroids (NEAs) are routinely discovered. 3,123 NEAs were discovered in 2024 alone1. Furthermore, new telescopes such as the Vera C. Rubin Observatory and the planned NEO Surveyor spacecraft are set to increase NEA discovery rates in the coming years (Mainzer et al., 2011; Schwamb et al., 2023). Once the orbit of a new object is determined, we must assess its Earth impact hazard. If a large enough asteroid is found to be on a collision course with the Earth, a deflection mission must be developed to mitigate the asteroid’s impact hazard. Thanks to NASA’s Double Asteroid Redirection Test (DART) mission, a Kinetic Impact (KI) is a demonstrated method of changing an asteroid’s trajectory (Chabot et al., 2024).Most asteroids that pose a significant impact hazard usually allow multiple impacts with the Earth over the span of decades. For example, Chesley et al. (2014) found that the orbit of asteroid (101955) Bennu allowed for more than 200 potential Earth impacts between 2167 and 2200. As it stands on 16 April 2025, more than 80% (29 out of 36) asteroids that have a Palermo asteroid impact hazard scale rating of -4 or greater admit more than one impact given their orbital uncertainties2. The fact that small variations in an asteroid’s orbit can cause future impacts has consequences for KI mission planning. Uncontrolled deflection can, for instance, unlock gravitational keyholes. Doing so can unlock a gravitational keyhole (Chodas, 1999). If pushed into a keyhole, an asteroid would return on an impacting trajectory. This means another deflection mission on possibly much shorter timescales could become necessary.In order to avoid such repetitive doomsday scenarios right off the bat, we should strive to design the initial KI deflection mission such that it does not trigger a keyhole (Chesley and Farnocchia, 2014; Eggl et al., 2018). In this work, we have developed a method that can aid KI mission decision-makers during the initial design process. We use realistic flyby mission trajectories that are already publicly available3 to obtain the KI velocity and mass at impact. We then combine this information with details of an impacting asteroid’s orbit, such as the keyhole locations and physical properties (like shape, spin state, and mass) and model hundreds of millions of KI mission outcomes. For each mission, we factor in realistic values of the momentum enhancement parameter as measured from the DART mission (Makadia et al., 2025).The change in the target asteroid’s velocity resulting from the KI mission is then mapped to the scattering encounter to assess whether a particular mission realization has the potential to trigger a keyhole. The impact probability of the asteroid following each simulated KI mission is then computed using the impact probabilities of each reachable keyhole. This information is convolved with realistic targeting uncertainties for a KI spacecraft. Finally, this process is repeated over one rotation phase of the asteroid to allow for additional control over the exact time of KI deflection.As a result of this process, we can map the 2014 keyholes of Bennu onto the surface of a target asteroid. Generating these impact probability maps allows KI mission designers to select the optimal site on the asteroid such that the post-deflection impact probability of the asteroid is minimized. Figure 1 shows a representative keyhole map on the surface of Bennu. The crosshair shows the optimal deflection site that minimizes the post-deflection impact probability of Bennu. By creating these maps, we can push asteroids away from the Earth such that they do not return on an impacting trajectory in the foreseeable future. We believe that this work is the next step in designing comprehensive KI deflection missions to ensure humanity’s continued safety from NEAs.Figure 1: Example post-deflection asteroid impact probability map on the surface of (101955) Bennu. The crosshair corresponds to the location on the surface that minimizes the asteroid impact hazard after deflection. These results assumed a 25-meter targeting uncertainty for the KI spacecraft. As a result, deflection sites that allowed for a KI miss are not considered and form a gray boundary around the targetable region of the asteroid.1https://cneos.jpl.nasa.gov/stats/totals.html2https://cneos.jpl.nasa.gov/sentry/3https://ssd.jpl.nasa.gov/tools/mdesign.html#/interactiveReferencesA. Mainzer et al. NEOWISE OBSERVATIONS OF NEAR-EARTH OBJECTS: PRELIMINARY RESULTS. The Astrophysical Journal, 743(2):156, December 2011. doi: 10.1088/0004-637x/743/2/156.M.E. Schwamb et al. Tuning the Legacy Survey of Space and Time (LSST) Observing Strategy for Solar System Science. The Astrophysical Journal Supplement Series, 266(2):22, May 2023. doi: 10.3847/1538-4365/acc173.N.L. Chabot et al. Achievement of the Planetary Defense Investigations of the Double Asteroid Redirection Test (DART) Mission. The Planetary Science Journal, 5(2):49, February 2024. doi: 10.3847/PSJ/ad16e6.S.R. Chesley et al. Orbit and bulk density of the OSIRIS-REx target Asteroid (101955) Bennu. Icarus, 235:5–22, June 2014. doi: 10.1016/j.icarus.2014.02.020.P.W. Chodas. Orbit uncertainties, keyholes, and collision probabilities. In Bulletin of the American Astronomical Society, volume 31, page 1117, January 1999. URL https://ui.adsabs.harvard.edu/abs/1999BAAS...31R1117C.S.R. Chesley and D. Farnocchia. Guided asteroid deflection by kinetic impact: Mapping keyholes to an asteroid’s surface. In Asteroids, Comets, Meteors 2014, page 92, July 2014. URL https://ui.adsabs.harvard.edu/abs/2014acm..conf...92C.S. Eggl, S.R. Chesley, P.W. Chodas, and D. Farnocchia. Avoiding Armageddon: Long-Term Asteroid Orbit Deflection Optimization. In 42nd COSPAR Scientific Assembly, volume 42, pages S.3–13–18, July 2018. URL https://ui.adsabs.harvard.edu/abs/2018cosp...42E.957E.R. Makadia, S.R. Chesley, et al. First detection of an asteroid’s heliocentric deflection: The Didymos system after DART. Submitted, 2025.