Pith. sign in

REVIEW 4 major objections 5 minor 29 references

TS-MAMP: A Remanufactured Agricultural Robot Powered by Second-Life EV Components and NMS-Free On-Device Weed Detection

T0 review · 4 major / 5 minor · reviewed 2026-08-04 · deepseek-v4-flash

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

Pith's one-line read This paper claims that retired low-speed EV motors and batteries, after modest screening, can be re-engineered into an affordable agricultural robot with an NMS-free weed detector running on edge hardware.

desk verdict A genuinely useful prototype-level systems integration paper with a good new dataset and a sensible NMS-free edge-training recipe, but the 60% affordability claim needs an auditable baseline before I'd trust the headline. read the letter →

arxiv 2608.02270 v1 pith:SPD4GYPW submitted 2026-08-03 cs.RO cs.AI

Weijie Shi , Zicheng Xu , Zhenbang Cheng , Haoran Xuan , Mingbo Duan , Gan Ge This is my paper · ORCID
classification cs.ROcs.AI
keywords agriculturalroboticsremanufacturingcirculareconomysecond-lifeEVcomponentsweeddetectionYOLOv10NMS-freesmallholderautomation
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

The paper sets out to prove that the growing stream of retired low-speed electric-vehicle parts can become the basis of an inexpensive agricultural robot rather than scrap. It reports a prototype, TS-MAMP, whose powertrain-and-chassis bill of materials comes to below USD 450—about 60% less than a new-component equivalent—by pairing salvaged 48 V hub motors with matched back-EMF signatures and reusing lead-acid modules that still hold 60–80% of rated capacity, actively balanced to within 100 mV. On the perception side it shows that an NMS-free YOLOv10n detector, one that skips non-maximum-suppression post-processing, reaches 80.87% mean average precision on a local five-class crop-weed dataset and runs on a Jetson Nano. The broader claim is that modest screening and modular integration can open a remanufacturing pathway for the fragmented smallholder fields that capital-intensive commercial automation ignores.

What carries the argument

The load-bearing pieces are: (1) back-EMF matching of recovered 48 V hub motors, which pairs waveforms and internal resistance so the two drive wheels behave symmetrically; (2) state-of-health screening of retired lead-acid modules at 60–80% capacity plus an active equalizer holding inter-module voltage within 100 mV, which makes heterogeneous second-life cells usable as a pack; (3) a telescopic-sleeve truss chassis with tool-change interfaces that converts dimension reconfiguration into affordability; and (4) an NMS-free YOLOv10n detector with a dual-assignment head, attention modules, and negative-sample training, which removes NMS post-processing so the perception stack can run determinis

What would settle it

Price out the new-component baseline in the same market with itemized supplier quotes; if two new 48 V/350 W BLDC motors, four new 12 V/20 Ah lead-acid batteries, controller, equalizer, frame steel, and chain drive total less than roughly USD 1,125, or if the retired-component build cannot be itemized under USD 450, the central cost claim is refuted. On the perception side, re-splitting the Wanxi Crop-Weed dataset and retraining/evaluating the same pipeline on a new 10% test split would show whether 80.87% mAP@0.5 holds or is a product of the particular split.

Watch

Extended reading notes

Core claim

The central claim is that retired EV components, under modest screening, can be re-engineered into an affordable AI-enabled agricultural robot. Concretely, the paper reports a prototype that pairs salvaged 48 V/350 W BLDC hub motors by back-EMF matching, assembles four second-life 12 V lead-acid modules into a 48 V pack with active balancing, and mounts them in a telescopic truss chassis that carries at least 200 kg, adjusts track width from 1200 mm to 2000 mm, and swaps tool modules in five minutes or less. On the perception side, the paper claims an NMS-free YOLOv10n detector reaches 80.87% mAP@0.5 on its own Wanxi Crop-Weed dataset and runs on a Jetson Nano in FP16 TensorRT, eliminating a

Load-bearing premise

The paper's affordability claim rests on the assumption that its 'equivalent new-component baseline' is a fair market price; no itemized quotes or sensitivity analysis are given, so if new 48 V BLDC motors, new lead-acid batteries, and a matching frame would cost materially less than assumed, the 60% saving and the below-USD 450 figure are overstated.

Editorial extensions

If this is right

  • A powertrain-and-chassis BOM below USD 450 would put a robotic field platform within reach of smallholders and cooperatives for whom commercial gantry or modular systems are out of budget.
  • Retired LSEV motors and batteries that currently go to destructive recycling could instead enter a screened, documented remanufacturing stream, cutting both cost and waste.
  • An NMS-free detector at 80.87% mAP@0.5 on a 675-image local dataset demonstrates that on-device crop-weed discrimination does not require a separate GPU or NMS implementation.
  • An adjustable, modular chassis allows one platform to serve multiple row spacings and tool heads, spreading the remanufacturing cost over several field operations.

Reading between the lines

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

  • The 60% saving is only as strong as the paper's implicit new-component baseline; an itemized, market-priced comparison in another country or at different scrap prices could shrink or enlarge the gap.
  • The 60–80% SOH screen is a capacity-versus-lifespan trade-off; a follow-up that tracks capacity fade under repeated field cycles would reveal whether the reused pack earns back its build cost before replacement.
  • The same screening-and-balancing recipe could generalize beyond LSEVs to e-bike or golf-cart drivetrains; the detector's current evidence, however, is limited to pak choi seedling stages in one region, so cross-crop and all-weather claims need their own validation.
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.

Referee Report

4 major / 5 minor

Summary. TS-MAMP is a modular agricultural mobile platform built from retired low-speed EV components: second-life BLDC hub motors matched by back-EMF, lead-acid modules screened at 60-80% SOH, and a telescopic truss chassis. The paper claims a ~60% powertrain/chassis BOM cost reduction (to below USD 450), mechanical specifications (≥200 kg load, 1200-2000 mm track width, ≤5 min changeover), and an NMS-free YOLOv10n weed detector reaching 80.87% mAP@0.5 on a locally collected crop-weed dataset, deployed on a Jetson Nano via TensorRT FP16. The central assertion is that modest screening of retired EV components yields an affordable AI-enabled agricultural robot for smallholder fields.

Significance. The sustainability angle is timely: second-life EV components are a real waste stream, and a working prototype with a sub-$450 BOM would be a meaningful demonstration. The paper also provides a useful ablation (Table II) showing monotone gains from augmentation, attention, and negative samples. However, the three headline claims—cost, mechanical specifications, and perception accuracy—are currently supported by assertions and one small test split rather than audit-grade measurements. The contribution is therefore promising but not yet demonstrated at the level required for a journal publication.

major comments (4)
  1. [Section II.A, Table I] The headline affordability claim ('approximately 60% lower', 'below USD 450') is presented without an auditable baseline. The remanufactured BOM is said to include 'transportation, screening labor, and fabrication overhead,' but the 'equivalent new-component baseline' is only described as 'commercially sourced 48 V BLDC motors, new lead-acid traction batteries, and a matching frame structure.' The text does not state whether the baseline includes the controller, battery equalizer, chain drivetrain, wiring, assembly labor, or overhead; nor are itemized prices, quantities, supplier quotes, or currency-year provided. Because the affordability claim is the central contribution, the paper must supply an itemized BOM comparison with explicit inclusion/exclusion criteria, unit costs, and sensitivity analysis (e.g., screening yield and labor rates). Without this, the 60% figure is an accounting
  2. [Section III.A-B, Table II] The perception results are based on a single 675-image dataset with a 65/25/10 random split, so the test set contains roughly 68 images and likely fewer than 100 instances per class. No per-class AP, no confidence intervals, and no repeated-split or cross-validation results are reported. Given the small test size, the 80.87% mAP@0.5 (58.41% mAP@0.5:0.95) and the ablation deltas in Table II may not be statistically stable. Please report per-class AP, number of test instances per class, and standard errors/bootstrap intervals, or use k-fold validation. Also report the deployment latency/FPS on the Jetson Nano; 'feasibility' is currently qualitative.
  3. [Sections II.B, II.D, III.C] Several mechanical and electrical specifications are asserted without measurement data: ≥200 kg static load, continuously adjustable track width 1200–2000 mm, ≤5 min module changeover, ±5 mm positioning accuracy, and inter-module voltage deviation below 100 mV during charge/discharge. The text says the load capacity was 'verified through physical prototype loading' and the equalizer 'maintains' the voltage differential, but no protocol, instrument, logged time series, or repeated-trial data are given. These specifications are load-bearing contribution claims; please provide measurement procedures and raw summaries (e.g., load-deflection curve, timed changeover trials, voltage deviation logs).
  4. [Section II.A] The 'modest screening' claim is not quantified. The battery window (60–80% SOH) was 'established through preliminary discharge characterization of 20 retired modules,' but the paper does not report the capacity distribution, the pass rate, or how many modules/motors were rejected. Screening yield directly affects BOM cost and the labor included in the USD 450 figure, so without yield data the cost calculation cannot be assessed. The paper also does not report how the back-EMF matching threshold was chosen or how many motors failed matching.
minor comments (5)
  1. [Section III.B] No inference speed or power-consumption figures are given for the Jetson Nano deployment. Add FPS, latency, and ideally power draw to substantiate 'on-device inference feasibility.'
  2. [Section II.C] The laser module is described with 50 W optical output and 'designed with reference to' GB/T 7247.1-2024; clarify whether the protective enclosure was actually tested/certified, since eye-safety compliance is load-bearing for the weeding module.
  3. [Section II.A] 'Preliminary discharge characterization of 20 retired modules' is a small sample; state whether this was a feasibility pre-study and provide the distribution of measured capacities.
  4. [References] References [1] and [7] are corporate/promotional sources; consider peer-reviewed alternatives for the NEXAT cost claim and the Naïo commercial claim.
  5. [Section III.B] The augmentation recipe ('offline dynamic mosaic and photometric augmentation') expands 439 images to 6,020; specify whether this was pre-generated or on-the-fly, and clarify that validation/test sets were untouched (the text says so, but the 'total training-time sample count' definition should be explicit).

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: measurements, ablations, and external citations; the cost baseline is an accounting comparison, not a fitted prediction.

full rationale

The paper is an empirical systems/benchmark report rather than a derivation. The central affordability claim is supported by a prototype BOM ledger (<USD 450) and a stated, though un-itemized, new-component baseline; the absence of itemized prices makes the 60% figure hard to audit but does not make it circular, since no quantity is defined in terms of a fitted parameter or of the conclusion. The battery SOH window (60–80%) is a screening criterion derived from preliminary discharge characterization of 20 retired modules, then applied to select modules; the resulting platform is tested, so the criterion is not being presented as a prediction from itself. Similarly, the NMS-free detector performance (80.87% mAP@0.5) is an ablation-measured outcome on a held-out test split, with YOLOv10 and Jetson-Nano benchmarks cited from external prior work; no self-citation is load-bearing and the authors cite no work of their own. The mechanical claims (≥200 kg static load, 1200–2000 mm track width, ≤5-min changeover) are physical validation results, not derived quantities. The only concern—the 'equivalent new-component baseline' for the 60% cost saving—is a comparison-standard transparency issue, not circular reasoning, and should be evaluated as an evidence-quality/correctness point rather than as circularity.

Assumptions & free parameters 4 free parameters · 4 assumptions · 2 invented entities

The central claims depend on hand-chosen screening thresholds and design parameters, plus several domain assumptions about retired-component reliability and dataset representativeness. No mathematical derivation is involved, so the main burden is empirical evidence that is not fully reported.

free parameters (4)
  • Battery SOH screening window = 60–80% SOH
    Selected after discharge characterization of only 20 retired modules; no distribution, capacity, or cycle-life data reported.
  • Active-balancing voltage deviation target = ≤100 mV
    Declared as the balancing threshold; no measured pre-/post-balancing voltage data or pack-level validation is given.
  • Negative-sample ratio for perception training = 500/6,520 (7.67%)
    Chosen to maximize final mAP; no sensitivity sweep or per-class breakdown is reported.
  • Chain reduction ratio = 3:1 (16T/48T)
    Design choice to compensate for aged-motor torque attenuation; no torque or efficiency measurements validate the margin.
assumptions (4)
  • domain assumption Retired lead-acid modules at 60–80% SOH retain sufficient capacity and cycle life for intermittent low-current agricultural duty.
    Invoked in Section II.A; supported only by 20-module preliminary characterization and cited literature, not by long-term tests in this paper.
  • domain assumption Back-EMF matching of salvaged BLDC hub motors preserves left/right kinematic symmetry.
    Section II.A states matching was performed, but no waveforms, resistance values, or symmetry measurements are reported.
  • domain assumption YOLOv10n's one-to-one head remains sufficiently accurate without NMS on this crop-weed domain.
    Borrowed from reference [24] and tested only on the authors' single-site dataset; no comparison to NMS-based detectors on the same data is provided.
  • domain assumption The 675-image Wanxi dataset from one site is representative enough for the reported mAP to generalize.
    Section III.A describes images from Lu'an only; no multi-site, multi-season, or external validation is presented.
invented entities (2)
  • Wanxi Crop-Weed Dataset
    purpose: Local training/evaluation corpus for the weed detection pipeline: 675 images, five classes.
    The dataset is described but not released and has no external link or independent benchmark comparison, so its validity and representativeness cannot be checked.
  • TS-MAMP telescopic-sleeve modular truss chassis independent evidence
    purpose: Reconfigurable load-bearing platform of the robot, with adjustable track width and module interfaces.
    The physical prototype is described with measurable specifications (≥200 kg, 1200–2000 mm track width, ≤5 min changeover), which an independent group could in principle verify by building or inspecting it.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TS-MAMP: A Remanufactured Agricultural Robot Powered by Second-Life EV Components and NMS-Free On-Device Weed Detection." pith.science (2026). https://pith.science/paper/SPD4GYPW

@misc{pith2026260802270,
  author       = {Pith},
  title        = {Pith review of: TS-MAMP: A Remanufactured Agricultural Robot Powered by Second-Life EV Components and NMS-Free On-Device Weed Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SPD4GYPW}},
  note         = {Machine review of arXiv:2608.02270}
}
read the original abstract

Agriculture 4.0 robotic systems improve field efficiency yet remain too capital-intensive for the fragmented smallholdings that dominate global agriculture. Meanwhile, a growing number of retired low-speed electric-vehicle (LSEV) powertrains retain functional electromechanical value but are destructively recycled. This paper presents TS-MAMP (Telescopic-Sleeve Modular Agricultural Mobile Platform), a remanufactured robot built under 3R (reduce, reuse, recycle) circular-economy principles. Retired 48 V brushless-DC (BLDC) hub motors are paired via back-EMF matching, and lead-acid battery modules screened at 60%-80% state of health are actively balanced within a 100 mV inter-module voltage deviation. Together, these reused components reduce the powertrain-and-chassis BOM cost by approximately 60%, to below USD 450 (perception and weeding modules excluded). The truss chassis provides >=200 kg static load, continuously adjustable track width from 1200 mm to 2000 mm, and <=5-minute module changeover. An NMS-free (non-maximum-suppression-free) YOLOv10n detector with consistent dual-assignment training and negative-sample learning achieves 80.87% mean average precision (mAP)@0.5 (58.41% mAP@0.5:0.95) on the Wanxi Crop-Weed dataset, and is deployed via FP16 TensorRT on a Jetson Nano, confirming on-device inference feasibility. TS-MAMP demonstrates that retired EV components, under modest screening, can be re-engineered into affordable, AI-enabled agricultural robots--opening a remanufacturing pathway for the smallholder fields that commercial automation leaves unserved.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

29 extracted references · 1 linked inside Pith

  1. [1]

    Unlocking potentials: Advantages of widespan controlled traffic farming compared to standard CTF,

    T. Chamen and J. McPhee, "Unlocking potentials: Advantages of widespan controlled traffic farming compared to standard CTF," NEXAT GmbH Corporate Technical Note, 2024

    2024

  2. [2]

    Agritechnica 2023: 1,100hp Nexat gantry system to harvest 200t/hour,

    O. Mark, "Agritechnica 2023: 1,100hp Nexat gantry system to harvest 200t/hour," Farmers Weekly, Nov. 2023

    2023

  3. [3]

    Theoretical study of transverse offsets of wide span tractor working implements and their influence on damage to row crops,

    V. Bulgakov, S. Pascuzzi, V. Adamchuk, V. Kuvachov, and L. Nozdrovicky, "Theoretical study of transverse offsets of wide span tractor working implements and their influence on damage to row crops," Agriculture, vol. 9, no. 7, p. 144, 2019

    2019

  4. [4]

    Achieving the sustainable agricultural development goals by adopting the new energy electric agricultural machinery: An analysis of opportunities and challenges of China,

    H. Yang et al., "Achieving the sustainable agricultural development goals by adopting the new energy electric agricultural machinery: An analysis of opportunities and challenges of China," Energies, vol. 18, no. 16, p. 4211, 2025

    2025

  5. [5]

    The Thorvald II agricultural robotic system,

    L. Grimstad and P. J. From, "The Thorvald II agricultural robotic system," Robotics, vol. 6, no. 4, p. 24, 2017

    2017

  6. [6]

    Thorvald II – a modular and re-configurable agricultural robot,

    L. Grimstad and P. J. From, "Thorvald II – a modular and re-configurable agricultural robot," IFAC-PapersOnLine, vol. 50, no. 1, pp. 4588–4593, 2017

    2017

  7. [7]

    Ted – Weeding Robot,

    Naïo Technologies, "Ted – Weeding Robot," 2026. [Online]. Available: https://www.naio-technologies.com/en/ted-robot/

    2026

  8. [8]

    Pathways to profit: Experimental evidence on agricultural technology adoption,

    FAO Investment Centre and Innovations for Poverty Action (IPA), "Pathways to profit: Experimental evidence on agricultural technology adoption," FAO Investment Centre Brief, Oct. 2023

    2023

Show all 29 references
  1. [9]

    Beyond the frontier: Decoding viability in smallholder finance,

    ISF Advisors, "Beyond the frontier: Decoding viability in smallholder finance," 2025 Rural and Agricultural Finance State of the Sector Report, Nov. 2025

    2025

  2. [10]

    Why smallholders (do not) adopt productivity-increasing technologies? Developing a smallholder adoption-process model through an in-depth qualitative study in Ethiopia,

    T. A. Abetu, P. T. M. Ingenbleek, K. E. Giller, E. Wolde-Meskel, and E. Baars, "Why smallholders (do not) adopt productivity-increasing technologies? Developing a smallholder adoption-process model through an in-depth qualitative study in Ethiopia," Technology in Society, vol....

    2026

  3. [11]

    China seeks to step up regulation of NEV battery recycling,

    Xinhua, "China seeks to step up regulation of NEV battery recycling," May 28, 2026

    2026

  4. [12]

    Repurposing second-life EV batteries to advance sustainable development: A comprehensive review,

    M. N. Akram and W. Abdul-Kader, "Repurposing second-life EV batteries to advance sustainable development: A comprehensive review," Batteries, vol. 10, no. 12, p. 452, 2024

    2024

  5. [13]

    Review of evaluation and decision-making framework for remanufacturing spent electric vehicle traction motors,

    E. Antwi, G. K. Ayetor, and F. K. Forson, "Review of evaluation and decision-making framework for remanufacturing spent electric vehicle traction motors," Preprints.org, 2026

    2026

  6. [14]

    Electromobility: Second life for electric motors,

    J. G. Erdmann, "Electromobility: Second life for electric motors," Fraunhofer-Gesellschaft Research News, Jan. 2024

    2024

  7. [15]

    A green strategy for upcycling utilization of core parts from end-of-life vehicles (ELVs): Pollution source analysis, technology flowchart, technology upgrade,

    R. Wang, L. Zhan, Z. Xu, R. Wang, and J. Wang, "A green strategy for upcycling utilization of core parts from end-of-life vehicles (ELVs): Pollution source analysis, technology flowchart, technology upgrade," Science of the Total Environment, vol. 912, p. 169609, 2024

    2024

  8. [16]

    Life cycle assessment of battery electric vehicles: Implications of future electricity mix and different battery end-of-life management,

    M. S. Koroma et al., "Life cycle assessment of battery electric vehicles: Implications of future electricity mix and different battery end-of-life management," Science of the Total Environment, vol. 831, p. 154859, 2022

    2022

  9. [17]

    Second life batteries lifespan: Rest of useful life and environmental analysis,

    L. C. Casals, B. A. García, and C. Canal, "Second life batteries lifespan: Rest of useful life and environmental analysis," Journal of Environmental Management, vol. 232, pp. 354–363, 2019

    2019

  10. [18]

    Exploring the state of health of electric vehicle batteries at end of use; hierarchical waste flow analysis to determine the recycling and reuse potential,

    N. Fallah and C. Fitzpatrick, "Exploring the state of health of electric vehicle batteries at end of use; hierarchical waste flow analysis to determine the recycling and reuse potential," Journal of Remanufacturing, vol. 14, no. 1, pp. 155–168, 2024

    2024

  11. [19]

    A survey of second-life batteries based on techno-economic perspective and applications-based analysis,

    H. Iqbal, S. Sarwar, D. Kirli, J. K. H. Shek, and A. E. Kiprakis, "A survey of second-life batteries based on techno-economic perspective and applications-based analysis," Carbon Neutrality, vol. 2, no. 1, p. 8, 2023

    2023

  12. [20]

    A review on circular economy: The expected transition to a balanced interplay of environmental and economic systems,

    P. Ghisellini, C. Cialani, and S. Ulgiati, "A review on circular economy: The expected transition to a balanced interplay of environmental and economic systems," Journal of Cleaner Production, vol. 114, pp. 11–32, 2016

    2016

  13. [21]

    Sustainable manufacturing: A review and framework derivation,

    V. M. Scharmer et al., "Sustainable manufacturing: A review and framework derivation," Sustainability, vol. 16, p. 119, 2024

    2024

  14. [22]

    A review of circular economy research for electric motors and the role of Industry 4.0 technologies,

    D. Tiwari, J. Miscandlon, A. Tiwari, and G. W. Jewell, "A review of circular economy research for electric motors and the role of Industry 4.0 technologies," Sustainability, vol. 13, no. 17, p. 9668, 2021

    2021

  15. [23]

    A circular economy approach for recycling electric motors in the end-of-life vehicles: A literature review,

    Z. Li, A. S. Hamidi, Z. Yan, A. Sattar, S. Hazra, J. Soulard, C. Guest, S. H. Ahmed, and F. Tailor, "A circular economy approach for recycling electric motors in the end-of-life vehicles: A literature review," Resources, Conservation and Recycling, vol. 205, p. 107582, 2024

    2024

  16. [24]

    YOLOv10: Real-time end-to-end object detection,

    A. Wang, H. Chen, L. Liu, K. Chen, Z. Lin, J. Han, and G. Ding, "YOLOv10: Real-time end-to-end object detection," in Advances in Neural Information Processing Systems 37 (NeurIPS 2024), 2024

    2024

  17. [25]

    GB/T 32620.1-2016, Lead-acid Batteries for Electric Road Vehicles—Part 1: Technical Conditions, AQSIQ/SAC, Beijing, China, 2016

    2016

  18. [26]

    Laser weeding with small autonomous vehicles: Friends or foes?

    C. Andreasen, K. Scholle, and M. Saberi, "Laser weeding with small autonomous vehicles: Friends or foes?" Frontiers in Agronomy, vol. 4, p. 841086, 2022

    2022

  19. [27]

    Design and testing of an autonomous laser weeding robot for strawberry fields based on DIN-LW-YOLO,

    P. Zhao, J. Chen, J. Li, J. Ning, Y. Chang, and S. Yang, "Design and testing of an autonomous laser weeding robot for strawberry fields based on DIN-LW-YOLO," Computers and Electronics in Agriculture, vol. 229, p. 109808, 2025

    2025

  20. [28]

    GB/T 7247.1-2024, Safety of Laser Products—Part 1: Equipment Classification and Requirements, State Administration for Market Regulation (SAMR), Beijing, China, 2024

    2024

  21. [29]

    Benchmarking deep learning models on NVIDIA Jetson Nano for real-time systems: An empirical investigation,

    T. P. Swaminathan, C. Silver, and T. Akilan, "Benchmarking deep learning models on NVIDIA Jetson Nano for real-time systems: An empirical investigation," arXiv preprint arXiv:2406.17749, 2024

    2024 arXiv

Pith tools

Reviewed August 4, 2026 · model on record in the stance chip above.