Andy Garcia · freelance dev
FRENES
Back to the blog
Space

Hera Saw Its Asteroid From 236,500 km as Less Than a Pixel. ESA Chose to Brake Later and Harder.

October 20, 2026 8 min read

On October 7, exactly two years after launch, the European Space Agency published the first image of Hera's destination taken by Hera itself. The Didymos binary asteroid system was about 236,500 kilometers away and smaller than a single pixel. The team found it by combining several images and running a photometric analysis to pick it out from the background stars.

That dot is two bodies: Didymos, about 780 meters across, and its moon Dimorphos, about 150 meters. On September 26, 2022, NASA's DART spacecraft hit Dimorphos on purpose. On October 11, 2022, NASA said the impact had shortened Dimorphos's 11-hour, 55-minute orbit by 32 minutes, give or take two; a 2023 paper in Nature by the DART team refined the change to 33 minutes, give or take one. Hera's job is the crash scene investigation: measure in detail how Dimorphos responded, so that DART becomes, in ESA's words, a "well-understood and potentially repeatable" planetary defence technique.

Eight days after ESA published that first glimpse, on October 15, Hera was scheduled to fire three of its thrusters for 93 minutes without a break. On October 16, ESA's official Bluesky account said the initial braking manoeuvre had been successfully completed, and ESA's operations account said the 93-minute burn had "performed within required parameters." ESA scheduled a second burn of about 31 minutes for October 22, then three small "transition" manoeuvres toward the end of October, and planned arrival for November.

I do not navigate spacecraft. I am CTO of a LegalTech company built on Claude. Hera raises a question that also comes up, at a far smaller scale, in the software I ship: how do you act on something you can only measure indirectly, and how much safety margin do you spend to go faster? ESA has published the plan and the team behind it, and its engineers described, in a 2024 paper, the more cautious design the agency has since compressed.

How small is "smaller than a pixel"

780 meters seen from 236,500 kilometers spans about 3.3 millionths of a radian, roughly 0.7 arcseconds. That is a one-euro coin seen from about seven kilometers. Hera did not see a shape, only a point of reflected sunlight among the stars.

In an April 2024 paper, ESA trajectory designers Pablo Muñoz and Michael Khan, with Ignacio Acedo of GMV, explain that Hera's approach starts from "very high phase angles": the spacecraft sees the asteroids mostly from their night side, which is "unfavourable for initial detection and optical navigation." In that design, the path bends toward the sunlit side, and the final approach spirals so that the direction of view keeps changing. A camera measures directions, not distances. Watching from changing angles is how directions become a position.

Three ways to locate a spacecraft

Before the asteroids appear in the camera, Hera is located by its radio signal, received on ESA's 35-meter deep space antennas. The signal's travel time gives its distance from Earth. Its Doppler shift, ESA writes, works "like a police radar gun" and gives the speed toward or away from Earth. Both measure along the line of sight; the position across the sky comes out only indirectly, and far less precisely. So the team periodically records the signal at two distant ground stations at once and measures the difference in its arrival times, calibrated against a quasar whose direction is known, a technique called delta-DOR. That reveals the sideways motion.

Once the asteroids fill the field of view, Hera tracks surface features as reference points. The instructive detail: the target is part of the estimate. Flight Dynamics engineer Michael Mueller lists what his team estimates besides Hera's own orbit: the asteroid system's orbit around the Sun, Dimorphos's orbit around Didymos, how both rotate, and their masses. You do not steer toward a known point. You steer toward a best guess that improves as you close in.

About 10 people work each shift, out of more than 50 assigned to Hera in ESA's Flight Dynamics team. Six teams look at the incoming data, "including sanity checks in the form of multiple independent assessments." Preparing a set of commands "might take 10 hours of detailed work at a time," Mueller says. ESA's Francesco Castellini explained that after the first braking burn, the team would use two kinds of evidence, Hera's own guidance sensor outputs and the ranging and Doppler data carried by its radio signal, to check that the direction and size of the speed change matched the plan. What the spacecraft reports, and what the radio link shows on its own.

Brake later and harder

On October 7, Hera was moving at more than 12 kilometers per second relative to Earth. That number matters less than it sounds: Hera and Didymos both orbit the Sun, so what has to be cancelled is the difference between them. The 93-minute burn was planned to remove "a few hundred metres per second" of it. If that means 200 to 400 m/s, Hera sheds 3.6 to 7 centimeters per second every second, less than one percent of the gravity you feel standing up.

In October 2025, ESA's Sylvain Lodiot said the reliability of Hera's propulsion let the team design "a new more aggressive manoeuvre plan that involves 'braking' later and harder," for an arrival a month earlier than first planned. The logic is simple: every day spent coasting fast covers more ground. Brake early, and you crawl the last stretch.

The 2024 paper shows what the slower plan was protecting. A single big burn, it says, "would not be robust": if it failed, Hera "would pass at high speed beyond the closest approach," and coming back would take "a long time and a delta-v penalty." So the pre-launch design used a test burn, then seven braking burns, one a week, each smaller than the last. The week between burns was for tracking, orbit determination and preparing the next command. The sizes were chosen so that a missed burn left time "to try to recover the spacecraft and command a new manoeuvre attempt" before closest approach. The whole rendezvous phase was to last two months.

The plan ESA described on October 7, 2026 keeps the skeleton and compresses it: a brief test burn on October 2, the 93-minute burn on October 15, then a burn of about 31 minutes a week later that should complete most of the speed change needed to match the asteroids, "at the same time compensating for any misperformance in the BRM-1." In March, ESA said the approach would last around three weeks.

ESA based the change on the propulsion's first-year record, and kept testing. The second deep-space manoeuvre, reported complete on March 17 after three burns and a small correction over about four weeks, used 123 kg of hydrazine for 367 m/s. Castellini said the team used it "to test all of the systems" needed for braking and rendezvous.

Choosing the failure mode

ESA estimates gravity at Didymos about 40,000 times weaker than Earth's, and about 200,000 times weaker at Dimorphos. Around the pair, Hera will move at a typical relative speed of about 12 centimeters per second, on a series of hyperbolic arcs first flown by Rosetta. ESA calls the technique "inherently safe because if any single manoeuvre fails, Hera will drift away from the asteroids rather than risk an impact."

The unknowns are real. Mueller points to "a lot of uncertainty, especially" about how much Dimorphos tumbles since DART's impact, says leftover impact debris "has not been entirely ruled out," and admits "we cannot simulate all probabilities." At comet 67P, Rosetta's star trackers were blinded by dust that showed up as false stars, though ESA notes Dimorphos is a less dusty asteroid.

What the evidence does not show

As of October 20, when I wrote this, ESA gave no arrival day, only November, and had not said how much recovery margin the compressed plan keeps. Its October 16 posts gave no speed-change figure for the first burn, and the second burn was still scheduled for October 22. The 2024 burn table was computed for a worst-case launch date, not the one Hera flew.

What I take from it

  1. Check an action through a channel other than the one that performed it. Hera's sensors say what the burn did; the radio signal says it independently. I do the same, more modestly, when I fact-check my posts with Jev as a judge instead of asking the model that wrote them.
  2. Spend margin you have measured. The 2024 design had seven weekly braking burns; the October 2026 plan had two, a week apart, and ESA credited the propulsion's reliability. It then tested the same systems again in March and on October 2.
  3. Keep a correction step inside the aggressive plan. The second burn was designed partly to absorb the first one's error.
  4. Pick the failure mode before you need it. An orbit that fails by drifting away is a design decision. In software, as a loose analogy, it is a job that stops cleanly instead of half-writing.
  5. Budget the cost of being sure: six teams, independent checks, up to 10 hours of work per set of commands.

ESA published Hera's first glimpse of its target eight days before the longest braking burn. Hera did not need to see more. It needed to know how wrong it might be, and to be wrong in the safe direction.

Sources

A project like this one?

I design and deploy products like this. Let's talk.

Let's talk