Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

RoboCulture: A Robotics Platform for Automated Biological Experimentation

T0 review · 3 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read A general-purpose robot arm with vision and force feedback autonomously ran a 15-hour yeast cell culture experiment, including the decision to split saturated wells.

desk verdict A solid integration paper with real engineering results, but the headline claim of autonomous sub-culturing is contradicted by the paper's own experimental account. read the letter →

arxiv 2505.14941 v2 pith:I2Q7GXZ7 submitted 2025-05-20 cs.RO

Kevin Angers , Kourosh Darvish , Naruki Yoshikawa , Sargol Okhovatian , Dawn Bannerman , Ilya Yakavets , Florian Shkurti , Alán Aspuru-Guzik
show 1 more author
Milica Radisic
This is my paper
classification cs.RO
keywords roboticlaboratoryautomationcellcultureself-drivinglabsvisualservoingopticaldensitymonitoringbehaviortreesyeastliquidhandling
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper introduces RoboCulture, a platform built around a general-purpose seven-axis robot arm, and claims it can carry out a complete cell-culture workflow without human intervention: pipetting liquids, replacing pipette tips, monitoring cell growth, and deciding when to split saturated cultures. The load-bearing demonstration is a 15-hour yeast experiment in which the robot filled wells with media, tracked growth by measuring the brightness of ordinary camera images, detected when cultures reached saturation, and split saturated wells into fresh wells, discarding contaminated tips between operations. The paper reports that the vision-guided pipette entered wells of a 96-well plate in random positions with 99% success, that force-guided tip attachment and removal succeeded in all 36 cycles tested, and that the camera-derived growth curves matched plate-reader curves from a parallel human-maintained culture. The underlying point is that flexible, relatively low-cost autonomy for biological experimentation may not require specialized liquid-handling hardware, only a general-purpose manipulator with closed-loop perception and a modular decision structure.

What carries the argument

The carrying mechanism is the behavior tree: a hierarchical, reactive controller that ticks reusable leaf behaviors and lets condition checks decide the order of operations, so the robot can pause, retry, or re-perceive when something fails. Underneath it sit three perception and control components. Image-based visual servoing computes the pixel error between the detected pipette tip and a target well, then commands incremental end-effector motions to drive that error to zero; this avoids depending on fragile hand-eye calibration, whose rotational error the paper shows grows linearly with camera-object distance. Force feedback performs tip exchange: the pipette body presses on a new tip while moving in a spiral, and the robot drops it in when the measured end-effector force falls, indicating that it has found the hole. Growth monitoring uses the brightness channel of small RGB images of each well, averaged over replicates and smoothed, as a relative optical-density signal; the robot reads the smoothed derivative to detect plateauing and trigger a split.

What would settle it

Dim the room lights by a fixed amount partway through a repeat of the 15-hour protocol and compare the robot's brightness curves with simultaneous plate-reader measurements of an identical plate; if the camera curves no longer track the plate reader or the split times shift, the lighting-stability assumption is falsified.

Watch

Extended reading notes

Core claim

On its own terms, the paper's discovery is that end-to-end execution of a standard cell-culture protocol is achievable with a general-purpose manipulator, vision, force feedback, and behavior trees, rather than with dedicated liquid-handling robots. RoboCulture converts the human sequence—seed, monitor, decide, split—into a hierarchy of reusable behaviors: image-based visual servoing steers the pipette tip to wells by minimizing pixel error instead of relying on calibrated world coordinates; a force-monitored spiral search turns pipette-tip attachment into a peg-in-hole task; and growth monitoring reads the value channel of RGB images to produce relative optical-density curves. The behavior tree runs a five-minute cycle of shaking, pausing, imaging well patches, smoothing brightness traces, and checking whether the growth curve is plateauing. When saturation is detected, the robot fills fresh wells with media, resuspends and aspirates the saturated culture, dispenses it into three daughter wells, voids the parent, and exchanges the tip. The paper reports that split times followed seeding density, that daughter wells showed growth patterns consistent with their parents, and that blank wells stayed negative, indicating no contamination.

Load-bearing premise

The saturation decisions are built on the assumption that lighting and reflections remain stable during the experiment, because growth is read from the brightness channel of ordinary camera images rather than from a true optical density measurement; if ambient light drifts, the robot could mis-time or miss a culture split.

Editorial extensions

If this is right

  • Cell-culture protocols that normally require overnight human attendance could be executed hands-free on flexible manipulator hardware.
  • Because the well plate can sit at arbitrary positions and the robot aligns by pixel error, lab layouts do not need to be rigidly jigged; only coarse fiducial offsets must be recalibrated.
  • The modular behavior-tree library means a new liquid-handling assay can be assembled from existing behaviors rather than reprogrammed as a new state machine.
  • Tracking relative growth by image brightness removes the need for a dedicated plate reader for split decisions, lowering the hardware cost of autonomous culture.
  • Interchangeable pipette tips with force-guided exchange allow long sterile protocols to run without cross-contamination between samples.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper does not test lighting robustness; a stress test with slowly drifting illumination would show whether the RGB-brightness growth signal, and therefore the split decisions, survive real laboratory conditions.
  • Extending the vision servo from the current XY plane to full six-degree-of-freedom control is the natural next step the paper itself flags, and would determine whether the same platform can handle tilted plates, viscous hydrogels, and organoid-scale pipetting.
  • Because the Digital Pipette v2 is described as costing under 300 USD and being assembled from printed parts and a linear actuator, the liquid-handling core of the system appears replicable; the real barrier for other labs is likely the manipulator and the calibration procedure, not the pipette.
  • The success criteria reported are engineering metrics; a biological assay with a measured endpoint would strengthen the claim that robot-performed splits produce cultures equivalent to human-performed ones.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper introduces RoboCulture, a laboratory automation platform built around a 7-axis Franka manipulator, an RGB-D camera, a custom open-source Digital Pipette v2, force-guided pipette tip exchange, and vision-based liquid handling into well plates. The authors report component-level validations: gravimetric pipetting accuracy against ISO 8655 and human pipettors, a 99% pipette insertion success rate into 96-well plates using image-based visual servoing, and 100% tip attachment/removal over 36 cycles. These components are integrated under a behavior-tree framework, and a 15-hour yeast culture experiment is presented in which the robot monitors growth via camera brightness, fills wells, performs splits, and exchanges tips. The paper claims a 'fully autonomous' experiment with 'decision-making for sub-culturing,' and releases code and CAD models.

Significance. If the component results hold, RoboCulture is a useful open-source contribution to flexible laboratory automation. The quantitative benchmarks against ISO standards and human operators, the closed-loop visual servoing approach to well-plate pipetting, and the force-based tip exchange are concrete strengths with reproducible supporting material. However, the paper's headline claim of fully autonomous sub-culturing decisions is not supported by the reported protocol: the manuscript itself states that splitting was manually initiated and that one group was manually instructed to split. The significance of the paper as a demonstration of end-to-end autonomy is therefore substantially reduced until the claim is revised or supported by additional evidence, such as an execution log showing that the robot's own saturation detection populated the split queue.

major comments (3)
  1. [Abstract; Section I (Contributions); Section II.F; Section III; Section IV.E] The headline claim that 'RoboCulture autonomously performs a 15-hour yeast cell culture experiment, integrating key tasks such as ... decision-making for sub-culturing' is contradicted by the paper's own protocol description. Section III states: 'While splitting was manually initiated as a safety precaution, RoboCulture accurately tracked the growth of the three replicates, determining the optimal splitting time based on the declining slope of the growth curve.' Section II.F further states: 'RoboCulture was manually instructed to split these wells pre-emptively due to time constraints' for the lowest-density group. Because the decision to split is represented by the needs_split parameter in the dynamic_reconfigure server, which can be set either by the behavior tree or by the human-operated rqt GUI (Section IV.E, Figure S16, Figure S1 caption), the reported experiment does not demonstrate autonomous sub-culturing decisions. To support the claim, please provide a run log showing that the robot's saturation-detection algorithm, rather than a human operator, populated the needs_split queue for the high- and intermediate-density groups; otherwise, the 'fully autonomous' and 'decision-making' wording should be removed or explicitly qualified as human-in-the-loop.
  2. [Section II.D; Figure 10] The growth-monitoring pipeline assumes lighting and reflections remain stable for the entire experiment, as stated in Section II.D: 'We assumed these conditions remain stable so that repeated measurements reliably mirror true growth trends without environmental interference.' Since all saturation detections and split timings derive from the HSV V channel, a lighting drift would cause mis-timed or missing splits. The validation in Figure 10 compares camera-derived curves with plate-reader curves from a separately prepared 'identical plate' (Section IV.F.2), not from the same wells; the curves are normalized and overlaid, but no correlation coefficient, per-group error, or statistical test is reported. The claim of 'reasonable agreement' therefore does not yet establish that the brightness signal reliably detects the plateau for split decisions. Please provide quantitative agreement metrics on the same plate, or explicitly frame the growth monitoring as qualitative and not validated for autonomous split timing.
  3. [Section III; Section V-B] The Discussion claims that the behavior tree enabled the robot to 'autonomously recover from errors and continue to operate without human intervention,' but no failure-injection experiments or recovery statistics are reported; the only evidence is anecdotal. If recovery from perception failures is part of the contribution, it should be quantified, for example by reporting the number of perception failures, detection latency, and successful recoveries observed during the 15-hour run. Otherwise, this sentence should be softened to a design rationale rather than a demonstrated capability.
minor comments (5)
  1. [Section II.B.1; Table I caption] The text refers to ISO 8655-2 for maximum permissible errors while the table caption cites ISO 8655-6; please make the standard references consistent.
  2. [Section II.E] There is a typo: 'RoboCulure' should be 'RoboCulture' at the end of Section II.E.
  3. [Figure 7; Section II.D] The y-axis label 'Value Channel Magnitude' is not self-explanatory; please state explicitly that it is the mean of the HSV V channel over the well image.
  4. [Figure 10; Section IV.F.2] The normalization procedure is described only for the camera-derived curves; please describe the normalization applied to the plate-reader curves and indicate whether any error bars or replicate variability are available.
  5. [Section IV.F.1] The component list after 'Figure 3shows the experimental setup' repeats the list already given in the Figure 3 caption; one of the two lists should be removed for conciseness.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the system's individual capabilities are validated against external ISO 8655 gravimetric tests, human pipettors, insertion trials, and plate-reader curves; the disclosed manual overrides affect the autonomy claim but are not a circular reduction.

full rationale

RoboCulture is an engineering demonstration rather than a fitted-prediction paper, so the main circularity patterns do not apply. The pipette's pulse-length-to-volume calibration (Section IV-B, Figs S12-S15) is an ordinary instrument calibration, and the reported accuracy is obtained afterward by the external ISO 8655-6 gravimetric procedure and by comparison with four human pipettors (Table I, Fig. 4), so the accuracy result is not an artifact of the calibration input. The visual-servoing controller's gain k_p and saturation limit u_lim (Algorithm 1) are engineering tuning parameters, and the claimed 99% insertion success is measured directly from randomized trials (Table II). The growth-monitoring pipeline uses the HSV V channel as a relative brightness proxy under an explicitly stated stability assumption (Section II.D) and is validated against plate-reader growth curves from a simultaneously prepared plate (Fig. 10), so no quantity is defined in terms of the result it is said to establish. Self-citations (e.g., the prior Digital Pipette, Ref. [14]) are present, but the load-bearing performance claims are externally benchmarked, so they are not circular. The substantive weakness is a claims-consistency issue, not circularity: Section II.F states that the lowest-density wells 'had not yet reached saturation. RoboCulture was manually instructed to split these wells pre-emptively due to time constraints,' and Section III states 'While splitting was manually initiated as a safety precaution,' which undercuts the abstract's phrase 'fully autonomous ... decision-making for sub-culturing.' This mismatch should be weighed as an overclaim or missing evidence about autonomy, but it does not make the derivation circular.

Assumptions & free parameters 3 free parameters · 5 assumptions · 2 invented entities

The central claim rests on engineering domain assumptions rather than on newly postulated physical entities. The only fitted quantities are hardware calibrations and manually chosen control/smoothing parameters, none of which are disguised as scientific predictions. The most fragile assumption is stable lighting for the optical-density proxy, followed by the sufficiency of planar visual servoing for the whole protocol.

free parameters (3)
  • Pipette pulse-length to volume calibration = Not reported numerically (Figures S12-S15)
    Section IV.B derives the relationship between actuator pulse length and delivered volume by calibrating around target volumes; this mapping is hardware calibration, not a scientific fit.
  • Visual servoing gains and limits (k_p, u_lim, img_thresh) = Not reported in text
    Algorithm 1 treats these as inputs to the feedback controller; their values affect insertion success and retry rate, but are not given.
  • Growth-curve smoothing parameters = Rolling average window = 5; baseline = first 10 data points
    Used in Section II.D and Figure 9 for saturation detection; these choices affect split timing and are not derived from a formal criterion.
assumptions (5)
  • domain assumption The well plate remains at a fixed height, so visual servoing in the XY plane with separately controlled Z is sufficient for pipetting.
    Invoked in Sections II.C and IV.C; restricts the system to planar pipetting and prevents generalization to tilted or 3D operations.
  • domain assumption Lighting conditions and reflections remain stable during the 15-hour experiment.
    Explicitly stated in Section II.D; the optical-density proxy is the V channel of RGB images, so lighting drift would corrupt growth curves.
  • domain assumption Object pose errors in the camera frame and robot encoder pose errors are negligible compared to hand-eye calibration error.
    Section V-D uses this to derive the error bound that motivates closed-loop servoing; it is not verified experimentally.
  • domain assumption FastSAM segmentation plus Canny edges plus a projective template transform yields well positions accurate enough for 9 mm wells.
    Section IV.C relies on this accuracy; the paper reports insertion success but not localization error separately.
  • domain assumption Behavior tree statuses and dynamic reconfigure state accurately model the experiment and its failure recovery.
    The 15-hour run depends on this software framework; no formal verification or systematic fault-injection results are provided.
invented entities (2)
  • Digital Pipette v2 independent evidence
    purpose: Replaceable-tip air-displacement pipette that a robot gripper can operate, for sterile liquid handling in the automated workflow.
    An open-hardware device with released CAD models, assembly instructions, and gravimetric test data, so independent groups can build and evaluate it; not a theoretical entity.
  • 3D-printed tip rack and tip remover independent evidence
    purpose: Store new pipette tips and detach used ones during autonomous tip exchange.
    CAD models are released (Figures S9-S10) and performance is measured in 36 cycles, so the hardware is reproducible.

how reviews work

0 comments
Cite this review

Pith. "Pith review of RoboCulture: A Robotics Platform for Automated Biological Experimentation." pith.science (2026). https://pith.science/paper/I2Q7GXZ7

@misc{pith2026250514941,
  author       = {Pith},
  title        = {Pith review of: RoboCulture: A Robotics Platform for Automated Biological Experimentation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/I2Q7GXZ7}},
  note         = {Machine review of arXiv:2505.14941}
}
read the original abstract

Automating biological experimentation remains challenging due to the need for millimeter-scale precision, long and multi-step experiments, and the dynamic nature of living systems. Current liquid handlers only partially automate workflows, requiring human intervention for plate loading, tip replacement, and calibration. Industrial solutions offer more automation but are costly and lack the flexibility needed in research settings. Meanwhile, research in autonomous robotics has yet to bridge the gap for long-duration, failure-sensitive biological experiments. We introduce RoboCulture, a cost-effective and flexible platform that uses a general-purpose robotic manipulator to automate key biological tasks. RoboCulture performs liquid handling, interacts with lab equipment, and leverages computer vision for real-time decisions using optical density-based growth monitoring. We demonstrate a fully autonomous 15-hour yeast culture experiment where RoboCulture uses vision and force feedback and a modular behavior tree framework to robustly execute, monitor, and manage experiments. Video demonstrations of RoboCulture can be found at https://ac-rad.github.io/roboculture.

Figures

Figures reproduced from arXiv: 2505.14941 by the authors.

Figure 1
Figure 1. RoboCulture integrates a vision-based liquid handling system capable of reliable pipetting into 96-well plates, a force￾guided pipette tip exchange system, and cellular growth monitoring toward generalizable biology laboratory automation. Biology protocols are represented as behavior trees, a reactive and modular framework for experiment state handling. Code, CAD models and video demonstrations of RoboCulture can be… view at source ↗
Figure 2
Figure 2. RoboCulture bridges human-level biological protocols and low-level robotic tasks using behavior trees. High-level experimental instructions such as growth monitoring and sub-culturing yeast are translated into modular, hierarchical behavior trees composed of individual behaviors. These trees coordinate key robotic subsystems, including optical density perception, vision-based robot control, and pipetting, enabling a… view at source ↗
Figure 3
Figure 3. The yeast growth experiment setup. 1) Our Digital Pipette v2, 2) a downwards-facing Intel RealSense D435i camera, 3) a Franka Emika robot with a Robotiq 2F-85 gripper, 4) an OHAUS SHHD1619DG Heavy Duty Orbital Shaker Platform, 5) a 3D printed pipette tip remover, 6) a biological waste bin, 7) a Falcon Tube rack holding YPD media, 8) a 3D printed pipette tip rack, and 9) a 96 well plate prepared with yeast. equipment… view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Gravimetric comparison of the Digital Pipette v2 with human pipettors, for volumes of 0.2 mL, 1 mL and 5 mL. A Levene’s test was conducted to compare the variances between the two groups (Digital Pipette v2 vs. Human). mL, and 5 mL. Each operator performed five replica…
Figure 5
Figure 5. Figure 5: Demonstration of the spiral search process for pipette tip attachment. Initially, the pipette body and the new tip are misaligned. A spiral trajectory of waypoints in the XY plane is generated around the start position of the pipette body, and the robot moves the pipet…
Figure 6
Figure 6. Figure 6: Overview of the perception pipeline. Raw images are obtained from the downward facing camera attached to the robot’s end effector. The raw images are preprocessed with a Gaussian blur, normalization and contrast adjustment. The pipette and the well plate are segmented …
Figure 7
Figure 7. Figure 7: Demonstration of well growth monitoring. RoboCulture captures images of the wells over time, extracting the magnitude of the value channel as a measure of image brightness. Changes in optical density are tracked to generate a growth curve. cases of suspended cell cultu…
Figure 8
Figure 8. Figure 8: The yeast growth experiment procedure. a) The configurations of the well plate before and after the experiment, showing the different initial concentrations and splitting pattern. Each group is split into the three rows below it on the well plate, and the original well…
Figure 9
Figure 9. Figure 9: Results from the yeast culture experiment. The y-axis shows image brightness, which correlates with the optical density of each sample. Colored solid lines represent the growth of three experimental groups with initial yeast concentrations of 50 million, 30 million, an…
Figure 10
Figure 10. Figure 10: Comparison of plate reader growth curves (dashed) with our growth curves obtained from the robot’s camera (solid). We implemented a behavior tree to enable reactive decision-making, allowing the system to adapt dynamically during execution. For instance, a perception …
Figure 11
Figure 11. Figure 11: (a) Photo of the assembled Digital Pipette v2. (b–e) 3D CAD models of the Digital Pipette v2 components: (b) platform, (c) syringe, (d) plunger, and (e) cover printing prices were estimated using online printing services (Xometry, Shapeways). Our pipette can be built …

Discussion (0). Sign in to comment.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. LabEvolver: Training-Free Experience Evolution for Safe and Grounded Wet-Lab Agents

    cs.RO 2026-07 conditional novelty 6.0 of 10

    A training-free dual-loop framework that distills completed lab-robot trials into reusable state-paired experience raises wet-lab pH-task efficiency by ~48% and ALFWorld Success@20 from 76.2% to 91.4%.

  2. GAMORA: A Gesture Articulated Meta Operative Robotic Arm for Hazardous Material Handling in Containment-Level Environments

    cs.RO 2025-06 reject novelty 4.0 of 10

    GAMORA is a VR-guided, ROS-controlled 3D-printed robotic arm that the authors report can place and pipette with roughly 2 mm accuracy, though the supporting measurements are not provided.

Reference graph

Works this paper leans on

39 extracted references · 28 canonical work pages · cited by 2 Pith papers

  1. [1]

    Self-driving laboratories for chemistry and materials science,

    G. Tom, S. P. Schmid, S. G. Baird, Y . Cao, K. Darvish, H. Hao, S. Lo, S. Pablo-Garc ´ıa, E. M. Rajaonson, M. Skreta, N. Yoshikawa, S. Corapi, G. D. Akkoc, F. Strieth-Kalthoff, M. Seifrid, and A. Aspuru-Guzik, “Self-driving laboratories for chemistry and materials science,”Chemical Reviews, vol. 124, no. 16, pp. 9633–9732, aug 2024. [Online]. Available: h...

    doi:10.1021/acs.chemrev.4c00055 2024

  2. [2]

    Next-generation experimentation with self-driving laboratories,

    F. H ¨ase, L. M. Roch, and A. Aspuru-Guzik, “Next-generation experimentation with self-driving laboratories,”Trends in Chemistry, vol. 1, no. 3, pp. 282–291, 2019. [Online]. Available: https: //doi.org/10.1016/j.trechm.2019.02.007

    doi:10.1016/j.trechm.2019.02.007 2019

  3. [3]

    Ot-2 liquid handling robot,

    O. L. Inc., “Ot-2 liquid handling robot,” 2024, accessed: 2024-11-11. [Online]. Available: https://opentrons.com/ robots/ot-2

    work page 2024

  4. [4]

    Microlab star v automated liquid handling platform,

    H. Company, “Microlab star v automated liquid handling platform,” 2024, accessed: 2024-11-11. [Online]. Available: https://www.hamiltoncompany.com/auto mated-liquid-handling/platforms/microlab-star-v

    work page 2024

  5. [5]

    Flowbot® one: The intuitive liquid handling robot,

    F. Robotics, “Flowbot® one: The intuitive liquid handling robot,” 2024, accessed: 2024-11-11. [Online]. Available: https://flow-robotics.com/products/flowbot-o ne/

    work page 2024

  6. [6]

    The stemcellfactory: A modular system integration for automated generation and expansion of human induced pluripotent stem cells,

    A. Elanzewet al., “The stemcellfactory: A modular system integration for automated generation and expansion of human induced pluripotent stem cells,” Frontiers in Bioengineering and Biotechnology, vol. 8, p. 580352, 2020. [Online]. Available: https://doi.org/10 .3389/fbioe.2020.580352

    work page arXiv 2020

  7. [7]

    Cell care star: Automated cell culture system,

    H. Company, “Cell care star: Automated cell culture system,” 2024, accessed: 2024-11-11. [Online]. Available: https://www.hamiltoncompany.com/automate d-liquid-handling/assay-ready-workstations/cell-care-s tar

    work page 2024

  8. [8]

    Automation in the life science research laboratory,

    I. Holland and J. A. Davies, “Automation in the life science research laboratory,”Frontiers in Bioengineering and Biotechnology, vol. 8, 2020. [Online]. Available: https://doi.org/10.3389/fbioe.2020.571777

    arXiv 2020

Show all 39 references
  1. [9]

    Accelerating discovery in natural science laboratories with ai and robotics: Perspectives and challenges from the 2024 ieee icra workshop, yokohama, japan,

    A. I. Cooper, P. Courtney, K. Darvish, M. Eckhoff, H. Fakhruldeen, A. Gabrielli, A. Garg, S. Haddadin, K. Harada, J. Hein, M. H ¨ubner, D. Knobbe, G. Pizzuto, F. Shkurti, R. Shrestha, K. Thurow, R. Vescovi, B. V ogel-Heuser, ´Ad´am Wolf, N. Yoshikawa, Y . Zeng, Z. Zhou, and H....

    2024 doi arXiv

  2. [10]

    Transforming science labs into automated factories of discovery,

    A. Angelopoulos, J. F. Cahoon, and R. Alterovitz, “Transforming science labs into automated factories of discovery,”Science Robotics, vol. 9, no. 95, p. eadm6991, 2024. [Online]. Available: https://doi.org/10 .1126/scirobotics.adm6991

    2024

  3. [11]

    Integrating a manual pipette into a collaborative robot manipulator for flexible liquid dispensing,

    J. Zhang, W. Wan, N. Tanaka, M. Fujita, and K. Harada, “Integrating a manual pipette into a collaborative robot manipulator for flexible liquid dispensing,” 2022. [Online]. Available: https://doi.org/10.1109/TASE.2023. 3312657

    2022 doi

  4. [12]

    A modular robotic platform for biological research: Cell culture automation and remote experimentation,

    J. Hamm, S. Lim, J. Park, J. Kang, I. Lee, Y . Lee, J. Kang, Y . Jo, J. Lee, S. Lee, M. C. Ratri, A. I. Brilian, S. Lee, S. Jeong, and K. Shin, “A modular robotic platform for biological research: Cell culture automation and remote experimentation,”Advanced Intelligent Systems...

    2024 doi

  5. [13]

    Foundation models in robotics: Applications,

    R. Firoozi, J. Tucker, S. Tian, A. Majumdar, J. Sun, W. Liu, Y . Zhu, S. Song, A. Kapoor, K. Hausman et al., “Foundation models in robotics: Applications,” Challenges, and the Future, 2023

    2023

  6. [14]

    Digital pipette: open hardware for liquid transfer in self-driving laboratories,

    N. Yoshikawa, K. Darvish, M. G. Vakili, A. Garg, and A. Aspuru-Guzik, “Digital pipette: open hardware for liquid transfer in self-driving laboratories,”Digital Discovery, vol. 2, pp. 1745–1751, 2023. [Online]. Available: http://doi.org/10.1039/D3DD00115F

    2023 doi

  7. [15]

    Colledanchise and P

    M. Colledanchise and P. Ogren,Behavior Trees in Robotics and AI: An Introduction, 07 2018

    2018

  8. [16]

    AprilTag: A robust and flexible visual fidu- cial system,

    E. Olson, “AprilTag: A robust and flexible visual fidu- cial system,” inProceedings of the IEEE International Conference on Robotics and Automation (ICRA). IEEE, May 2011, pp. 3400–3407

    2011

  9. [17]

    Piston- operated volumetric apparatus – part 2: Pipettes,

    International Organization for Standardization, “Piston- operated volumetric apparatus – part 2: Pipettes,” In- ternational Organization for Standardization, Tech. Rep. ISO 8655-2:2022(en), 2022

    2022

  10. [18]

    Piston-operated volumetric apparatus – part 6: Gravimetric reference measurement procedure for the determination of volume,

    ——, “Piston-operated volumetric apparatus – part 6: Gravimetric reference measurement procedure for the determination of volume,” International Organization for Standardization, Tech. Rep. ISO 8655-6:2022(en), 2022

    2022

  11. [19]

    S. I. A. I. Center and G. Agin,Real time control of a robot with a mobile camera, ser. Technical note. SRI International, 1979. [Online]. Available: 14 https://books.google.ca/books?id=XEAtGwAACAAJ

    1979

  12. [20]

    Robust estimation of bacterial cell count from optical density,

    J. Bealet al., “Robust estimation of bacterial cell count from optical density,”Communications Biology, vol. 3, no. 1, p. 512, 2020. [Online]. Available: https://doi.org/10.1038/s42003-020-01127-5

    2020 doi

  13. [21]

    Estimating microbial population data from optical density,

    P. Mira, P. Yeh, and B. G. Hall, “Estimating microbial population data from optical density,”PLOS ONE, vol. 17, no. 10, pp. 1–8, 10 2022. [Online]. Available: https://doi.org/10.1371/journal.pone.0276040

    2022 doi

  14. [22]

    Towards flexible biolaboratory automation: Container taxonomy-based, 3d-printed gripper fingers*,

    H. Zwirnmann, D. Knobbe, U. Culha, and S. Haddadin, “Towards flexible biolaboratory automation: Container taxonomy-based, 3d-printed gripper fingers*,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, Oct. 2023, p. 6823–6830. [Online]. A...

    2023

  15. [23]

    Precise well-plate placing utilizing contact during sliding with tactile-based pose estimation for laboratory automation,

    S. Pai, K. Takahashi, S. Masuda, N. Fukaya, K. Ya- mane, and A. Ummadisingu, “Precise well-plate placing utilizing contact during sliding with tactile-based pose estimation for laboratory automation,” in2024 IEEE/RSJ International Conference on Intelligent Robots and Sys- tems...

    2024

  16. [24]

    Lucidgrasp: Robotic framework for autonomous manipulation of laboratory equipment with different degrees of transparency via 6d pose estimation,

    M. Makarova, D. Trinitatova, Q. Liu, and D. Tsetserukou, “Lucidgrasp: Robotic framework for autonomous manipulation of laboratory equipment with different degrees of transparency via 6d pose estimation,” 2024. [Online]. Available: https: //doi.org/10.48550/arXiv.2410.07801

    2024 doi arXiv

  17. [25]

    Fast segment anything,

    X. Zhao, W. Ding, Y . An, Y . Du, T. Yu, M. Li, M. Tang, and J. Wang, “Fast segment anything,” 2023. [Online]. Available: https://doi.org/10.48550/arXiv.2306.12156

    2023 doi arXiv

  18. [26]

    A computational approach to edge detection,

    J. Canny, “A computational approach to edge detection,” IEEE Transactions on Pattern Analysis and Machine Intelligence, vol. PAMI-8, no. 6, pp. 679–698, 1986

    1986

  19. [27]

    The OpenCV Library,

    G. Bradski, “The OpenCV Library,”Dr . Dobb’s Journal of Software Tools, 2000

    2000

  20. [28]

    A modular robotic arm control stack for re- search: Franka-interface and frankapy,

    K. Zhang, M. Sharma, J. Liang, and O. Kroe- mer, “A modular robotic arm control stack for re- search: Franka-interface and frankapy,”arXiv preprint arXiv:2011.02398, 2020

    2011 arXiv

  21. [29]

    Sac- charomyces cerevisiae exponential growth kinetics in batch culture to analyze respiratory and fermentative metabolism,

    I. K. Olivares-Marin, J. C. Gonz ´alez-Hern´andez, C. Regalado-Gonzalez, and L. A. Madrigal-Perez, “Sac- charomyces cerevisiae exponential growth kinetics in batch culture to analyze respiratory and fermentative metabolism,”JoVE, no. 139, p. e58192, 2018

    2018

  22. [30]

    Automatic gener- ation and detection of highly reliable fiducial markers under occlusion,

    S. Garrido-Jurado, R. Mu ˜noz-Salinas, F. J. Madrid- Cuevas, and M. J. Mar ´ın-Jim´enez, “Automatic gener- ation and detection of highly reliable fiducial markers under occlusion,”Pattern Recognition, vol. 47, no. 6, pp. 2280–2292, 2014

    2014

  23. [31]

    Core processes in intelligent robotic lab assis- tants: Flexible liquid handling,

    D. Knobbe, H. Zwirnmann, M. Eckhoff, and S. Had- dadin, “Core processes in intelligent robotic lab assis- tants: Flexible liquid handling,” in2022 IEEE/RSJ Inter- national Conference on Intelligent Robots and Systems (IROS). IEEE, 2022

    2022

  24. [32]

    High-accuracy injection using a mo- bile manipulation robot for chemistry lab automation,

    A. Angelopoulos, M. Verber, C. McKinney, J. Cahoon, and R. Alterovitz, “High-accuracy injection using a mo- bile manipulation robot for chemistry lab automation,” in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2023, pp. 10 102–10 109

    2023

  25. [33]

    Vision-based robot manipulation of transparent liquid containers in a laboratory setting,

    D. Schober, R. G ¨uldenring, J. Love, and L. Nalpantidis, “Vision-based robot manipulation of transparent liquid containers in a laboratory setting,”arXiv preprint arXiv:2404.16529, 2024. [Online]. Available: https: //doi.org/10.48550/arXiv.2404.16529

    2024 doi arXiv

  26. [34]

    Precise dispensing of liquids using visual feedback,

    M. Kennedy, K. Queen, D. Thakur, K. Daniilidis, and V . Kumar, “Precise dispensing of liquids using visual feedback,” in2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2017, pp. 1260–1266

    2017

  27. [35]

    Preciseflex 3400 collaborative robot,

    B. Automation, “Preciseflex 3400 collaborative robot,” 2025, accessed: 2025-01-27. [Online]. Available: https: //www.brooks.com/industrial-automation/collaborative-r obots/preciseflex-3400/

    2025

  28. [36]

    Spinnaker microplate robot,

    T. F. Scientific, “Spinnaker microplate robot,” 2025, accessed: 2025-01-27. [Online]. Available: https://www. thermofisher.com/order/catalog/product/SPK0001

    2025

  29. [37]

    Openlh: Open liquid-handling system for creative experimentation with biology,

    G. Gome, J. Waksberg, A. Grishko, I. Y . Wald, and O. Zuckerman, “Openlh: Open liquid-handling system for creative experimentation with biology,” in Proceedings of the Thirteenth International Conference on Tangible, Embedded, and Embodied Interaction, ser. TEI ’19. New York, ...

    2019

  30. [38]

    An engineered approach to stem cell culture: Automating the decision process for real-time adaptive subculture of stem cells,

    D. F. E. Ker, L. E. Weiss, S. N. Junkers, M. Chen, Z. Yin, M. F. Sandbothe, S.-i. Huh, S. Eom, R. Bise, E. Osuna- Highley, T. Kanade, and P. G. Campbell, “An engineered approach to stem cell culture: Automating the decision process for real-time adaptive subculture of stem cel...

    2011 doi

  31. [39]

    move robot

    Description of Behavior Tree Components:Behavior trees are a hierarchical control architecture commonly used in robotics and game AI to model complex decision-making processes in a modular and reactive manner. A behavior tree is composed of nodes arranged in a tree structure, ...

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.