Force-Driven Validation for Collaborative Robotics in Automated Avionics Testing

David Frisini; Paolo Rocco; Pietro Dardano

REVIEW 4 major objections 5 minor 19 references

Force-Driven Validation for Collaborative Robotics in Automated Avionics Testing

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read The paper claims that a collaborative robot can validate its own cockpit interactions by classifying force, torque, and pose signals with convolutional neural networks, reaching F1-scores of about 95 to 98 percent on most avionics actions.

desk verdict A competent, honest feasibility study for force-based cobot action validation in a cockpit simulator; the central claim holds for the specific setup, and it deserves a serious referee despite known preprocessing tuning and label-quality caveats. read the letter →

arxiv 2505.10224 v1 pith:J4TGSBZH submitted 2025-05-15 cs.RO

Pietro Dardano , Paolo Rocco , David Frisini This is my paper · ORCID

classification cs.RO

keywords collaborativeroboticsdeeplearningexplainableAIforce-torquesensingconvolutionalneuralnetworksGrad-CAMavionicstestingsuccess/failureclassification

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 tries to show that the forces a collaborative robot feels while pressing buttons, flipping switches, and moving levers in an aircraft cockpit are enough to tell, on their own, whether each action succeeded. The authors record force, torque, and tool pose during over 4,330 manually labeled actions on an A320 simulator, preprocess the signals to remove sensor noise and controller oscillations, and train one-dimensional convolutional neural networks to classify each action as success or failure (or as a specific state transition). On five of the six tested components the networks reach F1-scores near or above 97 percent, with a knob-rotation task reaching about 95 percent after dataset augmentation; two-dimensional wavelet and hybrid variants perform slightly lower. The value of the claim, if true, is that automated certification testing can gain a self-contained, wrench-based check that runs in parallel with vision and is explainable through Grad-CAM heatmaps.

What carries the argument

The load-bearing mechanism is a multi-branch 1D-CNN whose inputs are cleaned by a three-stage preprocessing pipeline: (1) force/torque vectors are rotated from the robot base frame to the tool center point frame via $R^\top$; (2) a 30 Hz Butterworth filter suppresses sensor noise; and (3) a multi-energy-threshold transient detector picks the most informative 800-sample window of each signal, with thresholds at $0.38$, $0.22$, and $0.12$ of local maximum energy. Highly correlated signals are grouped into dedicated branches with customized kernel sizes and strides; GlobalMaxPooling follows the final convolutional layer, and dense layers map features to success/failure classes. For interpretability, 1D Grad-CAM computes gradient-weighted activations from the last convolutional layer, producing a temporal heatmap that shows which segments of force, torque, or pose drove the prediction. This whole assembly is what the paper argues turns a low-cost force sensor into a reliable validation instrument.

What would settle it

Take the trained 1D-CNN models and evaluate them on force, torque, and pose traces collected from the same actions on a real A320 cockpit rather than the simulator, without any retraining or preprocessing changes; a large F1 drop (for example, below 80%) would refute the claim that these models validate real avionics actions, while also revealing how much the result depends on simulator fidelity.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that 1D convolutional neural networks, trained on preprocessed force, torque, and pose signals, classify cobot actions on cockpit components with near-optimal F1-scores: 97.8% for buttons, 97.4% for switches, 97.5% for FLAP, 98.5% for landing gear, 98.0% for speed brake, and about 95% for knob rotation after addressing class imbalance. The preprocessing pipeline—transforming forces to the tool frame, a 30 Hz Butterworth filter, and an energy-based transient-isolation algorithm that selects an 800-sample window—is what makes the noisy, low-resolution embedded force sensor usable. Grad-CAM, applied in one dimension to preserve temporal structure, shows that the models focus on physically meaningful signal regions (for example, the final force spike of a landing-gear lever that indicates the lever was fully moved). The authors state these findings confirm the feasibility of the proposed wrench-based validation approach.

Load-bearing premise

The manual labels assigned to the recorded actions are treated as correct ground truth for success or failure, and the A320 simulator's force response is assumed to represent real aircraft cockpit components, so if either is wrong the reported F1-scores do not measure true validation accuracy.

Editorial extensions

If this is right

If the central claim holds, automated cockpit certification tests can be validated by the cobot's own force feedback, adding a redundant check alongside computer vision without extra hardware.
The 1D-CNN architecture becomes the recommended choice for this class of wrench-based validation: it matches or beats wavelet and hybrid models while training in under a minute on a consumer GPU.
The preprocessing steps (frame rotation, filtering, transient isolation) can be reused as a template for other contact-rich manipulation tasks where embedded force sensors are noisy.
Since the models are per-action and per-robot, scaling to other cockpit types or robot arms will require collecting new data; the paper's explicit next step is gathering data from diverse simulators and real cockpits.
Explainability via 1D Grad-CAM gives engineers a diagnostic tool: when a failure classification occurs, the heatmap shows which part of the signal triggered it, facilitating root-cause analysis.

Reading between the lines

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

The heuristic threshold values in the transient detector ($0.38$, $0.22$, $0.12$) are a likely source of setup-specific overfitting; a natural extension would be an adaptive or learned segmentation that tunes these thresholds per action type, and testing that would quantify how much of the reported F1 depends on them.
The same pipeline could be applied to other verification tasks where success is a discrete mechanical state change—automotive switchgear, industrial panel operation, medical device controls—by treating the force/torque trace as the ground-truth signature of the state transition.
The paper does not report cross-simulator or cross-platform transfer; a concrete test would be training on the A320 simulator and evaluating on a different cockpit simulator or a real cockpit to measure the domain gap in F1.
Because the knob task only reached its ~95% F1 after augmentation that may have induced data snooping, the paper itself flags this as the weakest result; a cleaner evaluation would collect more natural failures rather than augmenting the minority class.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

This is a straightforward feasibility study, and a fairly honest one. The new thing is the ARTO dataset: over 4330 labeled actions across six cockpit components, plus a per-action benchmark of 1D-CNN, 2D-CNN, and hybrid variants, and a 1D Grad-CAM adaptation that is genuinely useful for diagnosing where the model looks. The methods themselves are established—CNN on force signals, CWT scaleograms, Grad-CAM—but the domain application and the dataset are new, and the code is on GitHub. That counts for something.

The paper does several things well. The preprocessing pipeline is described in enough detail to reproduce, the authors explicitly admit when they tuned constants on test accuracy (the Butterworth cutoff, the energy thresholds), and they flag the knob augmentation as potential data-snooping. Section 6 is appropriately restrained: they say the algorithms are tailored to this robot and sensor, and that broader validation needs more data. That internal limitation statement matches what the evidence supports. The use of a held-out test split and per-action results is good practice.

Soft spots, in proportion. The labels are "assigned via script" and treated as ground truth, but there is no independent success criterion (e.g., switch state or electrical continuity) and no inter-rater or consistency check. That is the load-bearing premise, and it is only partially audited. The simulator fidelity question is real but not disqualifying for a feasibility claim—nobody expects an A320 simulator to match a real cockpit at this stage. The preprocessing tuning does introduce circularity, but the held-out split gives some independence, and the authors openly disclose the tuning. The lack of confidence intervals on the F1-scores is a minor issue; with per-action sample counts not given, it is hard to know how tight those 95–98% numbers actually are.

The citation pattern looks fine—relevant prior work is cited, and the authors do not overstate novelty. No invented entities, no hidden dependencies.

Who is this for? People working on force-based robot skill validation, automated aerospace testing, and applied interpretable ML. A reader will get a clear, reproducible baseline for this kind of wrench-based classification, together with an honest discussion of where the approach falls short.

Recommendation: send it to peer review. It is a solid engineering study with a public dataset and code, and the limitations are addressable in revision: add label consistency checks, report uncertainty, and clarify what the script-assigned labels actually measure. My verdict is conditional accept rather than accept—but it is a serious, worth-reading paper.

Referee Report

4 major / 5 minor

Summary. The paper proposes a force-driven validation framework for collaborative robots in automated avionics testing. The system uses a UR5e cobot with an embedded force-torque sensor to interact with cockpit components in an A320 simulator. Force, torque, and end-effector pose signals are recorded, preprocessed with a Butterworth filter and an energy-based transient detection algorithm, and then fed into 1D-CNN, 2D-CNN (using CWT scaleograms), and hybrid architectures. The models classify each action as Success or Fail and Grad-CAM provides visual explanations. Experiments on 4330 manually labeled actions across six components report F1-scores of roughly 95-98% for 1D-CNNs on most components, with the knob action requiring augmentation to reach 95.2%. The authors conclude that the approach is feasible for the specific robot configuration and call for more data from diverse simulators and real cockpits.

Significance. If the reported performance is trustworthy, the paper contributes an interpretable, wrench-based validation method that could reduce manual effort and increase reliability in avionics certification testing. The strength of the work lies in its breadth: it compares several network architectures, integrates 1D Grad-CAM for time-series interpretability, and involves a realistic A320 simulator setup. The authors also make the code available and are transparent about many limitations, including the risk of data-snooping in the knob augmentation. However, the evaluation is hampered by circularity in preprocessing selection and by an unclear labeling protocol, both of which directly affect whether the reported F1-scores reflect genuine functional validation. As a feasibility study the contribution is modest but plausible; the conclusions are currently stronger than the evidence supports.

major comments (4)

[Section 3 (Butterworth filter cutoff)] The Butterworth cutoff frequency (30 Hz) is selected by 'running CNN models and comparing their average classification accuracy' (Section 3). If this comparison uses the same test split from which the final F1-scores are reported, the evaluation is circular and the reported 95-98% scores are optimistically biased. The manuscript must describe the model-selection protocol (e.g., nested cross-validation, tuning on a validation subset, or a separate development set) and report the sensitivity of results to nearby cutoff values.
[Section 3 (ground-truth labels)] The dataset description is internally inconsistent: the text first says 'a label is assigned via script' and later refers to 'over 4330 manually labeled actions.' If the labels are assigned programmatically based on, say, encoder readings or commanded state, the paper should define the assignment rule and explain how it corresponds to functional success or failure (e.g., switch state, electrical continuity). If the labels are manually assigned, inter-rater reliability or an independent audit should be reported. Without a precise label definition, the F1-scores only measure agreement with the script/manual labels, not the validity of the action.
[Section 5.1 and Section 6 (Knob result inconsistency)] Table 1 reports a Knob F1-score of 95.2% after augmentation, while Section 6 states that 'the single exception is the knob task, where an unbalanced dataset constrains performance; however, an F1-score exceeding 78% is still achieved.' This discrepancy appears to reflect the authors' downweighting of the augmented result due to suspected data-snooping, but the paper never explicitly retracts or flags the Table 1 value in the conclusion. The contradictory claims must be reconciled: which result is the actual claimed outcome for the knob task, and what is the basis for trusting one over the other?
[Section 3 (energy-based transient thresholds)] The transient detection thresholds (0.38, 0.22, 0.12 of max energy) and the window lengths (300, 800 samples) are described as 'determined heuristically, posing a potential risk of overfitting to this specific setup and force controller settings.' Since these parameters are part of the feature extraction and were presumably chosen with knowledge of classification outcomes, they should be treated as hyperparameters. The paper does not report how varying these thresholds affects the final classification performance, nor does it provide an alternative, more principled selection criterion. At minimum, a sensitivity analysis is needed to gauge the robustness of the reported F1-scores to these choices.

minor comments (5)

[General (typos)] There are typographical errors, including 'traslation' in Section 3 and 'nternational' in reference [15]; these should be corrected.
[Table 1 (column heading)] The column heading 'Element' in Table 1 appears to denote the action type; please rename it to 'Action' for clarity.
[Section 5 (normalization details)] The description of when Standard Scaling is applied versus not is scattered; a concise summary of the normalization choice for each action and input channel would improve reproducibility.
[Grad-CAM evaluation] The Grad-CAM visualizations are presented qualitatively. Since interpretability is a stated contribution, the paper would benefit from a quantitative evaluation (e.g., localization metrics, perturbation tests, or expert agreement) to show that the highlighted regions correspond to meaningful action features.
[Code and data availability] The authors provide a GitHub URL, but it is unclear whether the dataset itself is included or only the code. Please clarify whether the data can be accessed for reproducibility.

Circularity Check

2 steps flagged · score 5.0 of 10

Reported F1-scores are partly fitted: the Butterworth cutoff is chosen by running CNN accuracy comparisons, and the paper itself flags the Knob augmentation as potential data-snooping; no external benchmark or independent success criterion is provided.

fitted input called prediction [Section 3, 'Setup and Data Preprocessing' (Butterworth filter cutoff)]
"Noisy signals are filtered using a Butterworth filter with a cutoff frequency (CF) of 30 Hz. This value was determined by running CNN models and comparing their average classification accuracy."

The preprocessing cutoff is selected by optimizing the same metric later reported as the main evidence, namely CNN classification accuracy/F1. As described, the paper does not state that this model-selection step used a held-out validation set disjoint from the final test set. The reported F1-scores therefore incorporate a parameter fitted to the evaluation target, so the final numbers are partly a fitted result rather than an independent estimate of the preprocessing-plus-classification pipeline.
fitted input called prediction [Section 5.1, '1D Convolutional Neural Networks' (Knob augmentation)]
"To address this, we augmented the dataset by up to half its size, focusing on less frequent cases, and employed extensive Dropout and MaxPooling. This approach yielded a final F1-score of approximately 95%, eliminating the previously observed overfitting. However, this significant improvement suggests potential data-snooping, reducing trust in the model compared to others."

The paper itself identifies the Knob result as potentially data-snooping. The augmentation was applied only after the initial imbalance and overfitting were observed, and the F1 improvement is attributed to that intervention. The 95% F1 for Knob is therefore not a clean out-of-sample measurement; it is the outcome of an iterative procedure informed by the observed performance, making that particular reported result partially circular by the authors' own admission.

full rationale

The paper's contribution is empirical rather than a formal derivation, and the held-out 60-20-20 / 70-30 splits provide some independence for the other action types. Still, two load-bearing evaluation choices are explicitly tied to the metric being reported: the 30 Hz Butterworth cutoff is chosen by comparing CNN classification accuracy, and the Knob augmentation is admitted as potential data-snooping. The transient-isolation thresholds are also acknowledged as heuristic with an overfitting risk. The claim that labels 'serve as ground truth for classification' is an external validity assumption about label correctness and simulator fidelity, not a constructional circularity, but it reinforces that no independent functional outcome criterion or external benchmark is used. The self-citations [1]-[2] are project-background references and are not load-bearing for the force-classification derivation. Overall, the central feasibility claim is partially circular because some of the reported high F1 values are fitted through preprocessing and augmentation choices guided by the same performance metric.

Assumptions & free parameters 5 free parameters · 5 assumptions · 0 invented entities

The paper contributes an empirical engineering result of applying established ML tools to a new robotics application. It relies on several data-dependent choices: the filter cutoff, transient thresholds, window lengths, channel selection, and CNN hyperparameters are all tuned using the same classification objective. The central assumptions are that manual labels are correct, that the simulator represents real cockpit interactions, and that force signals carry the failure signatures. No new physical entities or mathematical objects are introduced.

free parameters (5)

Butterworth filter cutoff frequency = 30 Hz
Selected by running CNN models and comparing average classification accuracy, meaning it was tuned against the target metric (Section 3).
Energy transient thresholds = 0.38, 0.22, 0.12 of max energy
Chosen heuristically for the setup; the paper states this "poses a potential risk of overfitting to this specific setup" (Section 3).
Transient window lengths = 300-sample energy window, 800-sample extraction window
Fixed lengths chosen for the setup without an independent selection criterion (Section 3).
Per-action signal channel selection = Selected versus All channels, varying per action
Chosen via PCA, Max-Energy analysis, and correlation matrices on the same data, affecting model input and results (Section 4).
CNN architecture hyperparameters = Ad hoc per branch; kernel sizes 20 to 100 samples, strides 10 to 50 percent of kernel
Tuned per action to improve accuracy and interpretability; exact values are not fully specified (Section 4).

assumptions (5)

domain assumption Manually assigned labels correctly represent functional success or failure.
Section 3 states the label "serves as ground truth for classification" for over 4330 actions, with no inter-rater reliability or external certified check described.
domain assumption The A320 simulator's force interactions are representative of real aircraft cockpit components.
The setup uses a simulator, while Section 6 projects eventual use in real cockpits; no cross-validation on real aircraft data is provided.
domain assumption Force, torque, and pose signals contain sufficient discriminative information for success versus failure classification.
This is the design premise of the whole approach; it is supported only indirectly by the reported F1-scores on the in-house dataset.
domain assumption The energy-based transient detection algorithm isolates the informative signal segment.
The algorithm and its thresholds are introduced in Section 3 and assumed to preserve failure-relevant content, though thresholds are tuned heuristically.
domain assumption The Butterworth filter and CWT processing do not remove failure-relevant information.
The 30 Hz cutoff is chosen by classifier accuracy, and the Morlet wavelet is selected for time-frequency balance; these choices are not independently validated.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Force-Driven Validation for Collaborative Robotics in Automated Avionics Testing." pith.science (2026). https://pith.science/paper/J4TGSBZH

@misc{pith2026250510224,
  author       = {Pith},
  title        = {Pith review of: Force-Driven Validation for Collaborative Robotics in Automated Avionics Testing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J4TGSBZH}},
  note         = {Machine review of arXiv:2505.10224}
}

read the original abstract

ARTO is a project combining collaborative robots (cobots) and Artificial Intelligence (AI) to automate functional test procedures for civilian and military aircraft certification. This paper proposes a Deep Learning (DL) and eXplainable AI (XAI) approach, equipping ARTO with interaction analysis capabilities to verify and validate the operations on cockpit components. During these interactions, forces, torques, and end effector poses are recorded and preprocessed to filter disturbances caused by low performance force controllers and embedded Force Torque Sensors (FTS). Convolutional Neural Networks (CNNs) then classify the cobot actions as Success or Fail, while also identifying and reporting the causes of failure. To improve interpretability, Grad CAM, an XAI technique for visual explanations, is integrated to provide insights into the models decision making process. This approach enhances the reliability and trustworthiness of the automated testing system, facilitating the diagnosis and rectification of errors that may arise during testing.

Figures

Figures reproduced from arXiv: 2505.10224 by the authors.

**Figure 4.** (a) Measured force feedback during multiple button pressing actions, high [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗

**Figure 5.** Comparative analysis of 2D and 1D Grad-CAM for feature attribution. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗

Figures from the paper (2 more)

**Figure 6.** Figure 6: Grad-CAM visualization: (a) The LDG lever is not pulled down with [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

**Figure 7.** Figure 7: Example of a 1D-GradCAM plot for the Speed Brake action, illustrating [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

19 extracted references · 13 canonical work pages

[1]

Frisini, D., Taumaturgo, V., et al.: Technology concept of an automated system for integration testing.CEAS Aeronautical Journal14, 1021–1030 (2023).https: //doi.org/10.1007/s13272-023-00709-3

work page doi:10.1007/s13272-023-00709-3 2023
[2]

https://doi.org/10.4050/F-0079-2023-18012

Frisini, D., Giulianini, G., et al.: Evaluation of an automatic system for cock- pit integration testing.79th Annual Forum of the Vertical Flight Society(2023). https://doi.org/10.4050/F-0079-2023-18012

work page doi:10.4050/f-0079-2023-18012 2023
[3]

Progress Report, European Commission, (2022)

AiRtificial Aerospace and Defense: ROSSI - Robotic System for In- spection. Progress Report, European Commission, (2022). Avail- able at:https://airtificial.com/wp-content/uploads/2022/09/ First-progress-report-for-dissemination.ab_-1.pdf

work page 2022
[4]

Avionics Magazine, (2019)

Lufthansa Technik AG: ROCCET - Robot Controlled Cockpit Elec- tronics Testing. Avionics Magazine, (2019). Available at:https: 12 Pietro Dardano et al. //interactive.aviationtoday.com/avionicsmagazine/may-2019/ disruptive-approaches-to-automated-test-equipment/

work page 2019
[5]

Eiband, T., Lee, D.: Identification of common force-based robot skills from the human and robot perspective.IEEE-RAS 20th Humanoids507–513 (2021).https: //doi.org/10.1109/HUMANOIDS47582.2021.9555681

arXiv 2021
[6]

https://doi.org/10.1038/s41467-022-32827-7

Zhang, J., Yao, J., et al.: Finger-inspired rigid-soft hybrid tactile sensor with su- perior sensitivity at high frequency.Nature Communications13(1), 5076 (2022). https://doi.org/10.1038/s41467-022-32827-7

work page doi:10.1038/s41467-022-32827-7 2022
[7]

2023.02.001

Castro, A., Baptista, J., Silva, F., Santos, V.: Classification of handover interac- tion primitives in a cobot–human context with a deep neural network.Journal of Manufacturing Systems68, 289–302 (2023).https://doi.org/10.1016/j.jmsy. 2023.02.001

doi:10.1016/j.jmsy 2023
[8]

Riffo, V., Pieringer, C., Flores, S., Carrasco, C.I.: Object recognition using tactile sensing in a robotic gripper.Insight - Non-Destructive Testing and Condition Mon- itoring64(7), 383–392 (2022).https://doi.org/10.1784/insi.2022.64.7.383

work page doi:10.1784/insi.2022.64.7.383 2022

Show all 19 references

[9]

In:IEEE International Conference on Robotics and Automation (ICRA), pp

Stolt, A., Linderoth, M., et al.: Detection of contact force transients in robotic assembly. In:IEEE International Conference on Robotics and Automation (ICRA), pp. 962–968 (2015).https://doi.org/10.1109/ICRA.2015.7139293

2015
[10]

https://doi.org/10.1109/LRA.2021.3072035

Tsuji,T.,Sato,K.,Sakaino,S.:ContactfeaturerecognitionbasedonMFCCofforce signals.IEEE Robotics and Automation Letters (RAL)6(3), 5153–5158 (2021). https://doi.org/10.1109/LRA.2021.3072035

2021
[11]

https://doi.org/10.1016/j.proeng.2012.01.843

Waghmare, R.G., Nalbalwar, S.L., Das, A.: Transient signal detection on the basis of energy and zero crossing detectors.Procedia Engineering30, 129–134 (2012). https://doi.org/10.1016/j.proeng.2012.01.843

2012 doi
[12]

Stathatos, E., Tzimas, E., Benardos, P., Vosniakos, G.-C.: Convolutional Neural Networks for Raw Signal Classification in CNC Turning Process Monitoring.Sen- sors24, 1390 (2024).https://doi.org/10.3390/s24051390

2024 doi
[13]

Russell, B., Han, J.: Jean Morlet and the continuous wavelet transform.CREWES Research Reports28, 1–12 (2016)

2016
[14]

Torrence, C., Compo, G.P.: A practical guide to wavelet analysis.Bulletin of the American Meteorological Society (BAMS)79(1), 61–78 (1998).https://doi.org/ 10.1175/1520-0477(1998)079<0061:APGTWA>2.0.CO;2

1998 doi
[15]

nternational Conferenceon Robotics and Automation (ICRA)8943-8950

Lee, M.A., Zhu, Y., Srinivasan, K., et al.: Making Sense of Vision and Touch: Self-Supervised Learning of Multimodal Representations for Contact-Rich Tasks. nternational Conferenceon Robotics and Automation (ICRA)8943-8950. (2019) https://doi.org/10.1109/ICRA.2019.8793485

2019
[16]

1007/s00170-017-1242-9

Chen, G., Zheng, Q.: Online chatter detection of end milling based on wavelet packet transform and SVM recursive feature elimination.International Journal of Advanced Manufacturing Technology95, 775–784 (2018).https://doi.org/10. 1007/s00170-017-1242-9

2018
[17]

Sun, Q., Yu, X., Li, H., Fan, J.: Adaptive feature extraction and fault diagnosis for three-phase inverter based on hybrid-CNN models.Neural Computing and Appli- cations33, 5111–5123 (2021).https://doi.org/10.1007/s40747-021-00337-6

2021 doi
[18]

2016.319

Zhou, B., Khosla, A., Lapedriza, À., Oliva, A., Torralba, A.: Learning Deep Fea- tures for Discriminative Localization.IEEE Conference on Computer Vision and Pattern Recognition (CVPR)2921–2929 (2015).https://doi.org/10.1109/CVPR. 2016.319

2015 doi
[19]

Shi, G., Wu, H., Luo, S., et al.: 1D gradient-weighted class activation mapping in spectroscopy analysis.Analytical Chemistry Journal (ACS)95(26), 9959–9966 (2023).https://doi.org/10.1021/acs.analchem.3c01101

2023 doi

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Frisini, D., Taumaturgo, V., et al.: Technology concept of an automated system for integration testing.CEAS Aeronautical Journal14, 1021–1030 (2023).https: //doi.org/10.1007/s13272-023-00709-3

work page doi:10.1007/s13272-023-00709-3 2023

[2] [2]

https://doi.org/10.4050/F-0079-2023-18012

Frisini, D., Giulianini, G., et al.: Evaluation of an automatic system for cock- pit integration testing.79th Annual Forum of the Vertical Flight Society(2023). https://doi.org/10.4050/F-0079-2023-18012

work page doi:10.4050/f-0079-2023-18012 2023

[3] [3]

Progress Report, European Commission, (2022)

AiRtificial Aerospace and Defense: ROSSI - Robotic System for In- spection. Progress Report, European Commission, (2022). Avail- able at:https://airtificial.com/wp-content/uploads/2022/09/ First-progress-report-for-dissemination.ab_-1.pdf

work page 2022

[4] [4]

Avionics Magazine, (2019)

Lufthansa Technik AG: ROCCET - Robot Controlled Cockpit Elec- tronics Testing. Avionics Magazine, (2019). Available at:https: 12 Pietro Dardano et al. //interactive.aviationtoday.com/avionicsmagazine/may-2019/ disruptive-approaches-to-automated-test-equipment/

work page 2019

[5] [5]

Eiband, T., Lee, D.: Identification of common force-based robot skills from the human and robot perspective.IEEE-RAS 20th Humanoids507–513 (2021).https: //doi.org/10.1109/HUMANOIDS47582.2021.9555681

arXiv 2021

[6] [6]

https://doi.org/10.1038/s41467-022-32827-7

Zhang, J., Yao, J., et al.: Finger-inspired rigid-soft hybrid tactile sensor with su- perior sensitivity at high frequency.Nature Communications13(1), 5076 (2022). https://doi.org/10.1038/s41467-022-32827-7

work page doi:10.1038/s41467-022-32827-7 2022

[7] [7]

2023.02.001

Castro, A., Baptista, J., Silva, F., Santos, V.: Classification of handover interac- tion primitives in a cobot–human context with a deep neural network.Journal of Manufacturing Systems68, 289–302 (2023).https://doi.org/10.1016/j.jmsy. 2023.02.001

doi:10.1016/j.jmsy 2023

[8] [8]

Riffo, V., Pieringer, C., Flores, S., Carrasco, C.I.: Object recognition using tactile sensing in a robotic gripper.Insight - Non-Destructive Testing and Condition Mon- itoring64(7), 383–392 (2022).https://doi.org/10.1784/insi.2022.64.7.383

work page doi:10.1784/insi.2022.64.7.383 2022