Integrity Review of “Design and control of a spine-like robotic arm actuated by shape memory alloys for high radiation environments”

Academic fraud report · Geng Detector

Summary

This review gives a “Questionable” verdict on the ChinaXiv preprint “Design and control of a spine-like robotic arm actuated by shape memory alloys for high radiation environments.” The strongest confirmed textual inconsistency is between the description of a “10-DOF” arm and Table 1, which lists 11 links (i = 1..11). Other confirmed or directly supported concerns include conflicting radiation-dose-rate statements of 40 rad/s (1440 Gy/h) and 1500 Gy/h, an undeclared potential funding–affiliation conflict, and a missing or ambiguous conflict-of-interest statement. Several additional issues remain unresolved. Table 2, Eq. 13, MOSFET measurements, and portions of the references appear corrupted or incomplete in the accessible PDF, but the available text cannot establish whether this occurred during PDF generation, extraction, or submission. Claims concerning a 110-fold stiffness change, shielding thresholds, tracking errors, and payload ratio also have plausible reporting or interpretation explanations. Automated image-forensic signals—including PRNU similarity, uneven noise variance, and copy-move matches—raise concerns, particularly for img-012, img-015, and img-019, but shared rendering pipelines, repeated graphical elements, or benign common sourcing cannot be excluded. Overall, the review identifies meaningful internal inconsistencies and requires source-data verification, but it does not provide definitive evidence of systematic fabrication or misconduct.

Verdict

Questionable. The review identifies one directly confirmed internal contradiction: the manuscript describes a “10-DOF” arm, whereas Table 1 lists 11 links (i = 1..11). Additional concerns include inconsistent radiation-dose-rate values, incomplete numerical content in Table 2 and Eq. 13, potentially misleading radiation and stiffness claims, reference-list problems, and an undeclared funding–affiliation relationship. Automated image-forensic results warrant source-image review, but they are not sufficient by themselves to establish image manipulation. The overall evidence supports further investigation rather than a finding of confirmed research misconduct.

Key findings

The paper is identified as a 2026 ChinaXiv preprint, ChinaXiv:202607.00209v1. The supplied data/video DOI is 10.57760/sciencedb.25498.
The manuscript states that the radiation chamber peaks at 40 rad/s (1440 Gy/h), but another passage reports a dose rate of 1500 Gy/h. On the stated conversion basis, this corresponds to approximately 41.67 rad/s.
Table 2 contains many cells with missing numbers, including entries such as “3.3° 0.°,” “11.° 1.°,” “17.° 0.°,” “3.° 8.°,” and “Average 2.° 6.° 2.°.” Exact reported errors therefore cannot be reliably verified.
Eq. 13 is partly illegible, with expressions including “θ(...SMA)= 0.6 sin(0.R_SMA+0.) + 0.” and “θ(...SMA)= 0.14R²_SMA − 0.R_SMA+ 0.” Fit coefficients and goodness-of-fit statistics cannot be confirmed.
The manuscript describes a stiffness change of up to 110 times, while separately referring to thresholds above 5 N·m² and below 2 N·m². The supplied evidence does not establish how the 110-fold figure was calculated.
The “10-DOF” description conflicts with Table 1, which contains 11 link rows, i = 1..11.
The paper states that the arm can hold 1500 g, described as more than 4.2 times its own weight of 360 g. Table 3 uses the labels “Weight 360 g” and “Load 1500 g,” but the interpretation is not fully explicit.
The manuscript reports MOSFET-related change rates of 8.8% and 12.1%, while also stating that irradiation had no significant effect on controller output. The underlying values could not be verified from the accessible text.
Image-forensic analysis produced PRNU and copy-move signals, including NCC 0.189, PCE 70.5, and same-offset match counts of 169, 165, and 124 for selected image pairs. These signals require raw-image and figure-production review.
Reference [31] appears to refer to an internal section, “Mechanism and theory,” rather than functioning as a complete external citation. Other entries contain apparent typographical errors, including “Thermocim. Acta” and “Mocroelectron. J.”
Funding is attributed to the “fund of innovation center of radiation application” (No. KFZC2021010301), while the authors are affiliated with the China Institute of Atomic Energy, Beijing 102413, China. A conflict-of-interest statement is not provided in the supplied report.

Evidence highlights

Confirmed textual contradiction: “For a 10-DOF spine-inspired flexible robotic arm” versus Table 1’s i = 1..11.
Dose-rate inconsistency: 40 rad/s (1440 Gy/h) appears in multiple passages, while another passage states dose rate: 1500 Gy/h.
Funding–affiliation link: “This work was supported by the fund of innovation center of radiation application (No. KFZC2021010301)” and “China Institute of Atomic Energy, Beijing 102413, China.”
Numerical corruption: Table 2 includes “3.3° 0.°,” “11.° 1.°,” “17.° 0.°,” “3.° 8.°,” and “Average 2.° 6.° 2.°.”
Image-forensic signals: I34 reports img-013 vs img-014 PRNU NCC=0.189, PCE=70.5; I17 reports img-012 copy-move (0,32) 169 pairs; I23 reports img-015 copy-move (32,0) 165 pairs; I29 reports img-019 copy-move (56,0) 124 pairs.
Noise variance signals: img-003 has noise CV 1.29, img-006 has 1.49, and img-008 has 1.23.
Incomplete fitting evidence: Eq. 13 is visibly corrupted and no reliable fit statistics are available in the cited text.

Notes

The accessible PDF appears to contain damaged or missing characters, particularly in numerical tables, equations, and some references. It is therefore not possible to determine whether these defects originated in the submitted document, PDF compilation, or text extraction. The image-forensic observations are computational indicators rather than definitive proof of manipulation; shared cameras, rendering or scanning pipelines, compression, and repeated graphical elements may produce similar signals. The reported “110-fold” stiffness claim, “4.2 times” payload ratio, shielding logic, and MOSFET interpretation should be checked against the original LaTeX files, raw data, radiation-dose records, figure files, and laboratory notebooks. The final determination of misconduct requires investigation by the relevant institution or publisher. The supplied DOI for the data/video is 10.57760/sciencedb.25498.