🤖 Build 10 · Automate the experiment

Automating the Experiment

A well-run AMO experiment looks after itself for most of the night. Servos hold powers, fields and frequencies. Scripts relock lasers and recalibrate whatever drifts. Optimisers tune one stage of the sequence while you sleep. This page is the practical guide: feedback and feedforward, beam-pointing stabilisation, relock and calibration loops, a workflow for optimising one part of the experiment, where machine learning has already been used, and what AI agents can and cannot be trusted with.

01 What to automate, and at which timescale

Automation is a stack of loops. Each one senses something, decides, and acts, and each closes at its own speed. The fast loops are electronics; the slower ones are scripts on the control PC; the slowest are optimisers and, since 2024, AI agents. A loop belongs at the slowest level that still keeps up with the disturbance it fights.

Layers of automation by timescale Bars on a logarithmic time axis from 100 nanoseconds to a day. Analog servos and real-time decisions act in under a millisecond; shot-to-shot feedback in seconds; relock and calibration loops in minutes to hours; optimisation runs in tens of minutes to days; AI agents over hours to days. 100 ns 1 µs 1 ms 1 s 1 min 1 h 1 day time for one pass round the loop (log scale) → Analog servoslock boxes · intensity servos · PDH Real-time decisionsFPGA: count photons, then act Shot-to-shot feedbackcorrect a drift using the last shot Relock and calibrationrelock, rescan a resonance, re-fit Optimisationscans, Nelder–Mead, Bayesian, M-LOOP AI agents (early)plan, run, interpret, report

Blue: electronics. Teal, amber and violet: software on the control PC. Bars show orders of magnitude for one pass round each loop, not limits. A PDH lock corrects in microseconds; a resonance recalibration takes a few minutes of shots; an optimisation run can take a night.

Automate first

What you do by hand every day

Relocking a laser, re-peaking a fibre, checking a resonance, re-measuring a π-pulse. These are well understood, easy to verify, and they eat hours of student time. Automating them also makes the data more consistent, because the check is done the same way every time.

Automate carefully

What can break hardware

Coil currents, high laser powers, RF amplifiers, vacuum valves and anything that heats. Limits for these belong in hardware (interlocks, current limits, thermal switches), with software limits on top. A loop that rails at full scale must fail safe, not try harder.

Not yet

What you do not understand

An optimiser maximises the number you give it, including its flaws. If the "atom number" can be inflated by a background count, a saturated camera or a mis-set threshold, the optimiser will find that first. Understand the measurement before you let a machine push on it.

02 Feedback and feedforward: holding a laser's power

The first servo most students build. A photodiode samples the beam after the AOM, a servo compares it with a setpoint and adjusts the AOM's RF amplitude. Feedback corrects whatever it sees, but only after it sees it and only up to its bandwidth. Feedforward acts before an error appears, but only on what you can predict or measure, and only as well as it is calibrated. Real systems use both.

Intensity servo with feedback and feedforward paths A laser beam passes a pick-off beam splitter, an AOM and a second sampling beam splitter before reaching the atoms. The sampling photodiode feeds a servo, which compares with a setpoint from the sequence and drives the AOM: the feedback path. The pick-off photodiode before the AOM drives the AOM directly: the feedforward path. Laser pick-off PD AOM sample PD atoms after the fibre, if there is one Servo setpoint r(t) from the sequence feedforward feedback

Power-servo simulator

50 ms of a beam whose laser power has mains ripple, slow drift, a sudden 8% drop at 20 ms and fast noise. The setpoint holds, then ramps up at 30 ms. Grey: what reaches the atoms with no control. Coloured: with your servo.

Servo

Simulated in the page at 0.25 µs steps. Laser power: 3% mains ripple, slow drift of about 2%, an 8% drop at 20 ms, 0.3% fast noise. Photodiodes: 0.1% noise. The AOM is a 0.5 µs first-order response after the loop delay, and its transmission is clamped between 0 and 1. The pick-off feedforward assumes a 5% calibration error. The numbers are illustrative; the behaviour is general.

Rules of thumb
  • Sample the light where the atoms are. After the fibre, if there is one; fibre coupling and polarisation drift change the power after the AOM.
  • The loop is only as quiet as its photodiode. Sensor noise is written onto the beam. Check with a second, out-of-loop photodiode: the in-loop one always looks good.
  • Delay sets the bandwidth. A pure integrator with a total delay τ goes unstable near f ≈ 1/(4τ), where the delay has used up its 90° of phase margin. Cables, the acoustic travel time in the AOM and digital processing all count.
  • Leave headroom. Run the AOM below full diffraction so the servo can push up when the laser sags. At the limit the servo saturates and the power simply follows the laser.
Practical details
  • Pulsed beams. A servo cannot regulate light that is off. Use a sample-and-hold (servo while on, hold the drive while off) or a digital loop that measures each pulse and corrects the next shot.
  • Which actuator. Feeding back to a diode laser's current also moves its frequency, so the AOM is the usual actuator. An EOM between polarisers ("noise eater") is faster still.
  • Setpoint feedforward is a calibration table: measured power against AOM drive. Then a ramp is followed immediately and the servo only cleans up the residual.
  • Digital or analog. FPGA servos (ARTIQ's SU-Servo combines ADC, servo and DDS amplitude in one loop) make the setpoint part of the sequence; analog PI boxes are simpler and have less delay.

03 Fields, frequencies and the mains

The same two ideas, feedback for what you cannot predict and feedforward for what you can, hold magnetic fields and laser frequencies steady. The mains is the textbook case for feedforward: its noise repeats every cycle.

Feedforward on the mains: a trapped-ion example

Merkel et al. stabilised a 14.6 mT field for a 43Ca+ clock qubit with a fluxgate sensor in a feedback loop, then added feedforward that cancels the 50 Hz mains harmonics. Qubit coherence time, from their Ramsey measurements:

Coherence time without stabilisation 0.082 ms, with feedback 0.57 ms, with feedback and feedforward 2.03 ms no stabilisation 0.082 ms · 112 nT feedback 0.57 ms · 16.1 nT feedback + feedforward 2.03 ms · 4.5 nT coherence time · rms field noise seen by the qubit

Values from Table II of Merkel et al., arXiv:1808.03310. Their abstract quotes 4.3 nT rms (0.29 ppm) for the stabilised field against more than 100 nT without stabilisation; the coherence time improved 25-fold.

What else is done this way
  • Line triggering. Start every shot at the same phase of the mains. The residual 50/60 Hz field is then the same in every shot, so it shifts a resonance by a fixed amount you can calibrate, instead of broadening it.
  • Laser frequency, fast. A PDH lock cannot suppress noise above its bandwidth. Chao et al. fed the PDH error signal forward to the light after a delay fibre and report phase-noise suppression around a few MHz about four orders of magnitude better than PDH feedback alone (Optica 2024). An adaptive version reports ≥40 dB around 2 MHz with up to 50 MHz bandwidth (preprint).
  • Laser frequency, slow. A software loop reads a wavemeter or a reference cavity and nudges a piezo or setpoint every second or so. It keeps a laser in range for the fast lock; see the laser locking guide.
  • Fields from known sources. If your own sequence switches a nearby coil, its stray field arrives at a known time. Feed the expected change forward to the compensation coils rather than waiting for the magnetometer to see it.

04 Keeping beams pointed: quadrant detectors and auto-alignment

Beams wander as the lab warms and cools, as mounts relax and as air moves. A position-sensitive detector turns that wander into an error signal and a piezo-driven mirror turns it back. The same hardware, with a smarter loop, can re-couple a fibre on its own.

A quadrant photodiode as an error signal

Drag the spot. The four quadrant signals give a horizontal error X = [(A+D) − (B+C)] / Σ and a vertical error Y = [(A+B) − (C+D)] / Σ, while Σ itself is the total power. Turn on drift to see the beam wander, then lock it.

For a Gaussian spot of 1/e² radius w on an ideal split detector, X = erf(√2·x/w): linear only for displacements well below w, then saturating at ±1. A small spot is sensitive but has a short range; a large spot the reverse. Detector size is in units of the half-width of the detector; gaps between quadrants are ignored.

Four degrees of freedom

A beam has a position and an angle in each transverse direction. One detector sees position; a second placed at the focus of a lens sees angle. Two detectors and two piezo mirrors close all four loops. One detector and one mirror only fixes the spot at the detector, and may tilt the beam elsewhere.

Commercial stabilisers

Closed-loop beam stabilisers (MRC Systems is a well-known example) combine piezo-driven steering mirrors with four-quadrant diodes and a controller. They are common in front of long beam paths, high-power fibres, SLMs and AODs, where a small angle becomes a large shift of a tweezer array.

Without a detector

When you only care about the throughput (a fibre), use the throughput itself: a photodiode after the fibre and motorised mirror mounts, driven by a search or an optimiser. Slower, but it optimises exactly what you need.

Automatic fibre coupling: three published approaches

ApproachHardwareMethodReported result
Model-based (Geng et al., 2024)Two 2D position-sensitive detectors, two piezo mirror mounts, a microcontrollerA linearised geometric model of the beam path, run as a state machineRecovers the maximum coupling in about 10 s, even from a beam misaligned to zero coupling preprint
Optimiser (Mathew et al., 2020)Stepper motors on the knobs of two kinematic mounts, a photodiode after the fibre, a Raspberry PiGaussian-process optimiser (M-LOOP)About 20 minutes for 200 runs; coupling as good as, or slightly better than, an experienced experimenter
Reinforcement learning (Richtmann et al., 2024)Motorised mirror mounts with backlash and lost stepsModel-free RL agent trained on the real setup90% coupling, comparable to a human expert; training took nearly four days at about one second per step preprint
The lesson generalises. If you can write a model of the system, a model-based loop is fastest. An optimiser needs no model but costs hundreds of measurements. Learning a policy costs the most, and pays off only when the task repeats under conditions a fixed model cannot capture.

05 Auto-relock and calibration loops

The most valuable automation in most labs is unglamorous: noticing that a lock has dropped, relocking it, and re-measuring the parameters that drift. Done well, it turns a machine that needs a student watching it into one that runs overnight.

Auto-relock state machine Locked: monitor the lock. When a lock-lost condition is seen, flag the shots, park the servo, scan and find the resonance, engage the lock, and verify it holds. If verified, return to locked. After three failed attempts, pause the experiment and alert a person. Lockedmonitor transmission,error signal, integrator Lost lockflag the shots, pausethe queue, park servo Searchscan piezo or current,find the resonance Engage and verifystep to the peak, close the loop,check it holds for N seconds verified: resume the queue failed: retry Three failures: stop and call a personleave the hardware in a safe state, send a message with the last scan attached The trigger condition and the"is it really locked?" test mattermore than the search itself.
Detecting a lost lock
  • Cavity transmission or a beat note drops below a threshold.
  • The error signal's rms grows, or the integrator output approaches its rail (it will soon run out of range even if still locked).
  • A wavemeter reading leaves its window.
  • The atoms tell you: fluorescence or loading rate collapses. Treat this as a symptom to diagnose, not as a lock signal on its own.
Make it honest
  • Flag, don't average. Every shot records the lock state of every laser; shots taken while anything was unlocked are marked bad, not silently averaged in.
  • Log every relock. A laser that relocks ten times a night is telling you something about its temperature or its current.
  • Give up gracefully. A relock script that keeps trying forever can drive a laser to a mode hop or a piezo to its limit.

Calibrations worth running on a schedule

What driftsCalibration experimentWhat the script updates
A transition frequency (field drift, light shifts)Short spectroscopy scan across the line, fit the centreThe detuning parameter used by every later experiment
Rabi frequency (beam power, pointing, polarisation)Rabi flop or a pulse-length scan, fit the periodThe π-pulse time or amplitude
Bias magnetic fieldMicrowave or Raman spectroscopy on a field-sensitive transitionCompensation-coil setpoints
Tweezer-to-tweezer uniformityPer-site light shift, loading or trap-frequency measurementPer-tone amplitudes on the AWG (see the tweezer designer)
Imaging thresholdHistogram of counts per site, fit the two peaksThe atom/no-atom threshold for each site
Calibrations depend on each other. A π-pulse calibration is meaningless if the resonance has moved. Google's superconducting-qubit group described organising calibrations as a directed acyclic graph: each node knows which others it depends on, checks them first, and only recalibrates what is out of date (Kelly et al., arXiv:1803.03226). The idea transfers directly to atoms. The cadence of each calibration should come from how fast that quantity actually drifts, which your monitoring logs (Section 10) will tell you.

06 Workflow: optimising one stage of the experiment

Say the tweezer-loading probability has dropped and you want to re-optimise the stage that loads atoms from the MOT into the tweezers. The steps below work for any stage, with or without machine learning.

  1. Define one number to maximise, measured at the end of the stage you are tuning (loading probability, atom number, temperature, survival). If you care about two things, combine them deliberately, for example atom number with a penalty above a temperature limit.Why: the optimiser will maximise exactly this number, flaws included (Section 01).
  2. Measure the noise of that number. Repeat the current settings 10–20 times. For a loading probability p from N shots the statistical error is √(p(1−p)/N): about 5% at p = 0.5 with 100 shots per point.Why: if two settings differ by less than this, no optimiser can tell them apart. It sets how many shots each point needs.
  3. Choose the parameters and safe bounds. Detunings, intensities, field gradients, durations. Start with the few you believe matter; bound each so that every combination is safe for the hardware.Why: cost grows with the number of parameters, and an optimiser will try the corners of the box.
  4. Freeze and monitor everything else. Lock states, laser powers, fields and the lab temperature should be logged alongside every shot.Why: a drift during the run looks to the optimiser like a change in the landscape.
  5. Look before you optimise. One-dimensional scans of each parameter around the current point show which ones matter and how smooth the landscape is.Why: if a 1D scan already finds the problem (a detuning off by a linewidth), you are done.
  6. Run the optimiser with reference points. Every so often re-measure the original settings.Why: if the reference drifts, the experiment changed under you, and the run must be corrected or restarted.
  7. Validate the winner. Re-measure the best settings many times, interleaved with the old ones. The best single measurement in a noisy run is biased high (the "winner's curse").Why: the improvement you report should survive a fair comparison.
  8. Understand it. Look at what the optimiser changed, and use its model of the landscape (if it has one) to see which parameters are sensitive.Why: a surprising optimum often reveals a misaligned beam or a wrong calibration, which is worth more than the gain itself.
  9. Write it down and schedule it. Commit the new values with a note (see control systems: git). If the optimum drifts, turn the stage into a periodic calibration (Section 05).Why: next month's student needs to know what changed and why.

07 Choosing an optimiser

Every shot costs seconds and every measurement is noisy, so the question is not which optimiser is cleverest but which finds a good setting in the fewest shots, without being fooled by noise. The race below runs four common strategies on the same toy landscape.

Optimiser race on a noisy two-parameter landscape

Each panel shows the true landscape (stronger colour is better) and the points each strategy measured. The chart below shows, after each shot, how good the strategy's current best guess really is. Raise the noise to see which methods get fooled.

Grid scanRandomNelder–MeadBayesian (Gaussian process)

A toy landscape, not a real MOT: a broad main peak, a smaller second peak and a gentle slope, with Gaussian measurement noise in units of the peak height. Grid, random and Nelder–Mead report the best point they measured; the Bayesian optimiser reports the measured point its model rates highest, which is what protects it from lucky noisy shots. The grid uses the largest square number of shots that fits the budget. Results vary with the noise seed, so press "New noise" a few times before drawing conclusions.

Scans

Grid and 1D scans

Easy to understand and to plot, and the right first look. The cost is kd points for k values of d parameters, so a grid is hopeless beyond two or three parameters.

Simplex

Nelder–Mead

Needs no model and few points per step. It was the workhorse of early online optimisation, but it can mistake noise for a slope and stall on a ridge, and it only finds the nearest peak.

Model-based

Bayesian optimisation

Fits a Gaussian process (a smooth model with error bars) to every shot so far, then measures where the model says improvement is likely. Efficient for expensive, noisy shots and tens of parameters; Vendeiro et al. tuned up to 55.

Population

Evolutionary methods

Differential evolution and CMA-ES keep a population of settings and breed better ones. Robust to noise and local peaks, but they spend more shots; good when shots are fast.

Neural networks

Neural-network learners

Replace the Gaussian process with a network when there are many parameters. Tranter et al. tuned a 63-parameter MOT compression sequence this way.

Policies

Reinforcement learning

For a different task: not one best set of numbers, but a policy that reacts to what it observes during the shot. Costly to train; usually trained in simulation first.

Tools. M-LOOP (Machine-Learner Online Optimization Package, from the group behind Wigley et al.) is built for exactly this job: it offers Gaussian-process, neural-net, differential-evolution, Nelder–Mead and random learners and talks to an experiment through files or a Python interface. General libraries such as scikit-optimize or BoTorch work too if you write the interface yourself. Either way the optimiser only proposes settings; your control system runs them (see control systems).

08 Machine learning in AMO labs: what has been done

Machine learning entered cold-atom labs as an optimiser for sequences too long to tune by hand, then spread to readout, control and alignment. These are representative papers, with what each one actually reports.

PaperWhat was automatedMethodReported result
Wigley et al., Sci. Rep. 6, 25890 (2016)Evaporation ramp for a Bose–Einstein condensateGaussian-process online optimiser (the origin of M-LOOP)A ramp producing high-quality BECs in 10 times fewer iterations than the previous online optimiser (Nelder–Mead)
Tranter et al., Nat. Commun. 9, 4360 (2018)MOT compression sequence for maximum optical depthStochastic artificial neural network; 63 parameters (21 time bins × 3 controls)Optical depth 970 ± 20 against 535 ± 8 for the best human-designed sequence
Barker et al., Mach. Learn.: Sci. Technol. 1, 015007 (2020)Laser cooling and evaporation together, for a BECDifferential evolution, Gaussian process and neural network comparedFour times the BEC atom number of the manually optimised settings; a BEC from fully randomised starting parameters
Vendeiro et al., Phys. Rev. Research 4, 043216 (2022)Fast BEC production via Raman cooling in an optical trapBayesian optimisation of up to 55 control parametersBECs of 2.8 × 10³ 87Rb atoms from a room-temperature gas in 575 ms
Reinschmidt et al., Nat. Commun. 15, 8532 (2024)Live control of a 87Rb MOT (detuning and field gradient) from fluorescence imagesDeep reinforcement learning, trained in simulation and transferred to the experimentAdaptive loading control; compensated a sudden detuning step and met atom-number targets on demand
Phuttitarn et al., Phys. Rev. Applied 22, 024011 (2024)State readout of a Cs tweezer arrayConvolutional neural networks on the camera imagesUp to 56% (single-site) and 32% (whole-array) lower readout error; crosstalk correlation reduced by up to 78.5%
Lee & Chae, arXiv:2401.04893Choosing tweezer moves for defect-free rearrangementReinforcement learning (PPO), simulations of 50–300 tweezersA simulation study, compared against A* path search; no experiment preprint
Reading these papers. Notice what they compare against. "Better than our manual settings" depends on how hard the humans tried; "fewer shots than Nelder–Mead" is a cleaner benchmark. The lasting value is often in what the optimiser found: Tranter et al.'s sequences and Vendeiro et al.'s fast Raman-cooling route are now physics, not just numbers.

09 AI agents and language models in the lab

Since late 2024 several groups have connected large language models to real experiments, letting them plan, run and interpret measurements. These are early results, mostly preprints, and the architecture of each system is as instructive as the result.

Superconducting qubits

k-agents (Cao et al.)

LLM agents turn a lab procedure written in plain language into a state machine of experiments, run it, and use separate inspection agents to judge each result before moving on. They ran calibrations autonomously for hours on a 16-qubit processor, reaching 99.87% single-qubit gate fidelity after recalibration.

arXiv:2412.07978; journal version in Patterns (2025).

Cold atoms

QCopilot (Sha et al.)

A team of LLM agents (decision maker, experimenter, analyst, fault diagnoser, recorder) ran Bayesian optimisation on a Rb cold-atom apparatus and diagnosed faults. The authors report 108 sub-µK atoms without human intervention within a few hours, about 100 times faster than manual tuning.

arXiv:2508.05421 preprint

NV centres

Agentic quantum sensing (Isogawa et al.)

An LLM agent planned and interpreted NV-centre measurements, deciding for itself to run a CPMG follow-up after a weak Ramsey feature. It had no direct hardware access: it wrote requests to a queue that deterministic control software checked and executed.

arXiv:2607.25145 preprint

Where LLMs help today
  • Writing and explaining instrument drivers, analysis scripts and sequence code, which you then test like a new student's code.
  • Reading the night's logs and summarising what failed and when.
  • Turning a procedure you already do by hand into a script or a checklist.
  • Searching manuals and datasheets for the setting you need.
Guardrails
  • Hardware interlocks stay independent of all software, agent or not.
  • The agent proposes, the control system disposes. Requests go through a queue that checks every parameter against fixed, safe ranges (the design in the NV-centre paper above).
  • A person approves anything irreversible or hazardous: high laser power, coil currents, valves, bake-outs.
  • Every action is logged with the reasoning the agent gave, so a bad night can be reconstructed.

10 The infrastructure that makes automation safe

Automation without monitoring is a machine that fails quietly. These four pieces are what let a group trust an experiment that runs while nobody is watching.

Log every slow variable

Laser powers, lock states, wavemeter readings, temperatures and humidity, vacuum pressures, coil currents: sampled continuously into a time-series database with dashboards (a common choice is InfluxDB with Grafana). When the atom number dropped at 3 a.m., the logs show that the lab temperature rose at 2:40.

Alert a person

A message to the group chat or a phone when a lock fails repeatedly, a pressure rises, cooling water stops, or a calibration lands far from its usual value. Alerts should be rare and specific, or people learn to ignore them.

One source of truth for parameters

Calibrated values (resonances, π-times, thresholds) live in one database or file with history. Every shot records which values it used, and every automatic update is logged with the data that justified it.

Test it without atoms

Run new automation against a simulator or with the lasers blocked before letting it loose. Check what it does when a sensor is unplugged, a reading is NaN or a device does not answer: those are the failures that happen at night.

11 Checklist before you let it run overnight

  • Every loop has limits on its output and a defined safe state when it gives up.
  • Every shot records the lock states, the parameters used and the code version.
  • The figure of merit has been checked by hand: it cannot be raised by a background, a saturated camera or a mis-set threshold.
  • Reference measurements are interleaved so drift can be told apart from improvement.
  • Someone gets a message if it stops, and knows what to do.
  • In the morning, validate before you believe: re-measure the winner against the old settings.
Next in the build: coupling atoms to a cavity. The control systems page covers the hardware these loops run on, and the lab handbook the hands-on alignment that comes before any of this.

12 References and further reading

Feedback, feedforward and stabilisation

B. Merkel et al., "Magnetic field stabilization system for atomic physics experiments," arXiv:1808.03310. Fluxgate feedback plus mains feedforward; the numbers in Section 03.
Y.-X. Chao et al., "Pound-Drever-Hall feedforward: laser phase noise suppression beyond feedback," Optica 11, 945 (2024), arXiv:2309.09759.
Y.-X. Chao et al., "Robust high-frequency laser phase noise suppression by adaptive Pound-Drever-Hall feedforward," arXiv:2407.19642 preprint
MRC Systems beam position stabiliser, Laser Focus World product note (2010): piezo steering mirrors with four-quadrant diodes.

Auto-alignment

Y. Geng et al., "A compact and open-source microcontroller-based rapid auto-alignment system," arXiv:2405.02306 preprint
R. S. Mathew et al., "The Raspberry Pi auto-aligner: machine learning for automated alignment of laser beams," arXiv:2010.15246.
L. Richtmann et al., "Model-free reinforcement learning with noisy actions for automated experimental control in optics," arXiv:2405.15421 preprint

Calibration and optimisation

J. Kelly et al., "Physical qubit calibration on a directed acyclic graph," arXiv:1803.03226 preprint
P. B. Wigley et al., "Fast machine-learning online optimization of ultra-cold-atom experiments," Sci. Rep. 6, 25890 (2016); M-LOOP documentation.
A. D. Tranter et al., "Multiparameter optimisation of a magneto-optical trap using deep learning," Nat. Commun. 9, 4360 (2018).
A. J. Barker et al., "Applying machine learning optimization methods to the production of a quantum gas," Mach. Learn.: Sci. Technol. 1, 015007 (2020).
Z. Vendeiro et al., "Machine-learning-accelerated Bose-Einstein condensation," Phys. Rev. Research 4, 043216 (2022).
M. Reinschmidt et al., "Reinforcement learning in cold atom experiments," Nat. Commun. 15, 8532 (2024).
L. Phuttitarn et al., "Enhanced measurement of neutral atom qubits with machine learning," Phys. Rev. Applied 22, 024011 (2024).
Y. Lee and E. Chae, "Machine learning-enhanced optical tweezers for defect-free rearrangement," arXiv:2401.04893 preprint

AI agents

S. Cao et al., "Agents for self-driving laboratories applied to quantum computing," arXiv:2412.07978; journal version "Automating quantum computing laboratory experiments with an agent-based AI framework," Patterns 6 (2025).
R. Sha et al., "LLM-based multi-agent copilot for quantum sensor," arXiv:2508.05421 preprint
T. Isogawa et al., "Agentic AI for scientific reasoning in autonomous quantum sensing experiments," arXiv:2607.25145 preprint

See Also