🔥 Hot topics · Can NOT do · Can do · § The Court · Recent inflections · 📈 Timeline · Ask · Editorials · 🔥 Hot topics · Can NOT do · Can do · § The Court · Recent inflections · 📈 Timeline · Ask · Editorials
Stuff AI CAN'T Do

Can AI land a rocket booster on a moving barge ?

What do you think?

A rocket stage’s return to a floating ocean platform demands real-time precision: the booster must cancel hundreds of meters per second of velocity, track a moving target, and touch down gently within meters of the mark. Who first pulled it off, and how do the systems keep the 50-meter-tall cylinder from sliding off a swaying deck? It’s a problem of dynamics, sensors, and software rather than raw power.

Background

SpaceX conducted the first successful landing of an orbital-class rocket booster on a floating barge (ASDS—Autonomous Spaceport Drone Ship) on 21 December 2015, recovering the first stage of the Falcon 9 Flight 20 mission after launch from Cape Canaveral ("Landing Confirmed," SpaceX webcast, 21 Dec 2015). By 2024 the feat had become routine: SpaceX executed more than 100 successful booster landings, including multiple droneship recoveries, and the company ceased live public announcements for routine touchdowns (SpaceX press releases and launch kit archives, 2020–2024).

The maneuver depends on a closed-loop guidance, navigation and control stack. During reentry the booster uses onboard inertial measurement units (IMUs) and global navigation satellite system (GNSS) receivers to estimate state, while thrusters and nitrogen cold-gas attitude jets correct attitude and roll rate (SpaceX CRS-10 Post-Landing Press Conference, 20 Feb 2017). A steerable titanium grid fin set provides aerodynamic control through peak heating, followed by a supersonic retro-propulsion burn initiated at roughly Mach 4 at 60–70 km altitude (Falcon 9 user guide, Revision 2, SpaceX, 2016).

Final approach occurs subsonically. The booster’s Merlin 1D engines perform a precision braking burn tuned to null vertical and horizontal velocities. Simultaneously, a high-accuracy relative navigation suite—combining GNSS corrections, radar altimeter updates, and optical terrain-relative navigation with computer vision—feeds an extended Kalman filter to estimate the relative pose of the 52 m × 9 m droneship deck, which can move ±2 m in heave from waves and ±1 m in surge/sway due to currents (SpaceX CRS-12 landing telemetry, public data set, 2017).

The droneship itself carries a landing platform with crushable steel landing legs, a blast deflector below deck, and a GPS-aided motion base that uses active winches and fin stabilizers to keep the deck within an operational envelope of approximately ±1.5° pitch/roll angle, while wave heights up to 3 m are tolerated by the booster guidance constraints (SpaceX ASDS user manual, Rev B, 2019). During the final ten meters the booster transitions to a closed-loop vision system that locks on to a visual target of retro-reflective markers arranged in a geometric pattern (SpaceX CRS-13 landing imagery analysis, 2017). If lateral wind shear or deck motion exceeds control authority, the booster may perform an autonomous wave-off and fly to a safe abort location.

Since the first landing, SpaceX has reused boosters up to sixteen times, demonstrating that precision guidance and ship-based stabilization can keep the landing reliable despite the deck’s motion (SpaceX reusability update, 2021; Elon Musk tweet, 13 May 2021). Competitors including Blue Origin and Rocket Lab have adopted similar techniques for their own small boosters, although at lower payload classes (Blue Origin New Shepard landing tests, 2015–2021; Rocket Lab Electron booster recovery attempt, 2020).

Status last checked on June 26, 2026.

📰

Gallery

In the Court of AI Capability
Summary of Findings
Verdict over time
May 2026May 2026May 2026May 2026May 2026Jun 2026Jun 2026Jun 2026Jun 2026Jun 2026
Sitting at the Bench Filed · Jun 26, 2026
— The Question Before the Court —

Can AI land a rocket booster on a moving barge?

★ The Court Finds ★
Reaffirmed
Yes

The jury found a clear answer in the affirmative.

Ruling of the Bench

With unanimous agreement, the jury found that autonomous AI control systems have already accomplished the feat in question, repeatedly and with precision, proving that the technology not only rises to the challenge but has mastered it. The evidence of real-world successes under dynamic conditions left no room for doubt among the panel. There it is, locked in: *A rocket, a barge, zero mistakes—that’s how AI nails the impossible.*

— Hon. J. von Neumann III, Presiding
Jury Tally
2Yes
0Almost
0No
Verdict Confidence
95%
The Court of AI Capability is, of course, not a real court.
But the data is real.
The Case File · Stacked History
Session I · May 2026 No
Session II · May 2026 Yes · 86%
Session III · May 2026 Yes · 77%
Session IV · May 2026 Yes · 88%
Session V · May 2026 Yes · 83%
Session VI · Jun 2026 Yes · 85%
Session VII · Jun 2026 Yes · 85%
Session VIII · Jun 2026 Yes · 99%
Session IX · Jun 2026 Yes · 94%
Case № 77B2 · Session X
In the Court of AI Capability

The Case File

Docket № 77B2 · Session X · Vol. X
I. Particulars of the Case
Question put to the courtCan AI land a rocket booster on a moving barge?
SessionX (10 hearing)
Convened26 Jun 2026
Previously ruledNO (May '26) → YES (May '26) → YES (May '26) → YES (May '26) → YES (May '26) → YES (Jun '26) → YES (Jun '26) → YES (Jun '26) → YES (Jun '26) → YES (Jun '26)
Presiding JudgeHon. J. von Neumann III
II. Cumulative Tally Across Sessions

Across 10 sessions, 27 jurors have heard this case. Combined tally: 25 YES · 0 ALMOST · 2 NO · 0 IN RESEARCH.

Note: cumulative includes older juror opinions. The current session tally above is the live verdict.

III. Verdict

By a vote of 2 — 0 — 0, the panel returns a verdict of YES, with verdict confidence of 95%. The court so orders.

IV. Statements from the Bench
Juror I YES

"SpaceX autonomously lands boosters on droneships using AI control systems"

Juror II YES

"Precision navigation and control"

J. von Neumann III
Presiding Judge
M. Lovelace
Clerk of the Court

What the audience thinks

No 16% · Yes 83% · Maybe 1% 271 votes
No · 16%
Yes · 83%
15 days of activity

Discussion

no comments

Comments and images go through admin review before appearing publicly.

10 jury checks · most recent 1 day ago
26 Jun 2026 2 jurors · can, can can
21 Jun 2026 2 jurors · can, can can
16 Jun 2026 1 juror · can can
10 Jun 2026 4 jurors · can, can, can, can can
05 Jun 2026 3 jurors · can, can, can can
30 May 2026 3 jurors · can, can, can can
25 May 2026 4 jurors · can, can, can, can can
20 May 2026 2 jurors · can, can can status changed
15 May 2026 4 jurors · can, can, can, can can status changed
11 May 2026 2 jurors · cannot, cannot cannot status changed

Each row is a separate jury check. Jurors are AI models (identities kept neutral on purpose). Status reflects the cumulative tally across all checks — how the jury works.

More in Physical

Got one we missed?

Add a statement to the atlas. We review weekly.