Learning a System-Level Surrogate for Hydraulic Excavators: A Simulation-to-Real LSTM Approach

Chenyu Wang; Donghai Shi; Muguo Du; Qiang Wang; Shen Wang; Shuai Wang; Xiaofeng Tao

REVIEW 3 major objections 5 minor 17 references

Learning a System-Level Surrogate for Hydraulic Excavators: A Simulation-to-Real LSTM Approach

T0 review · 3 major / 5 minor · reviewed 2026-08-01 · deepseek-v4-flash

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

Pith's one-line read A data-driven LSTM surrogate can stand in for both simulator and physical hydraulic excavator under identical control inputs.

desk verdict Solid engineering paper with a real validated result on a full-size excavator, but the 'drop-in replacement for the physical machine' claim outruns the evidence, mostly because evaluation ground truth is the authors' own Kalman-filtered signal. read the letter →

arxiv 2607.15656 v1 pith:RAVFQSBU submitted 2026-07-17 cs.RO cs.SYeess.SY

Shuai Wang , Shen Wang , Qiang Wang , Muguo Du , Donghai Shi , Chenyu Wang , Xiaofeng Tao This is my paper · ORCID

classification cs.ROcs.SYeess.SY MSC 68T0793E1193C85

keywords hydraulicexcavatordigitalsurrogateLSTMsystemidentificationsim-to-realtransferKalmanfilteringclosed-loopbehavioralequivalenceautoregressiverollout

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 establish that a whole hydraulic excavator, with its unmodeled hydraulic dynamics, delays, and sensor noise, can be replaced by a learned input-output operator trained end-to-end from sensor data. The authors argue that an LSTM network, fed with recent joint states and control commands and driven autoregressively, reproduces the machine's angular velocities and long-horizon joint-angle trajectories well enough to serve as a functional stand-in for both a MuJoCo simulator and a real SY750H excavator. They introduce a consistency-aware adaptive Kalman filter to reconcile inconsistent raw angle and angular-velocity measurements before training, and they report high angle-trajectory R-squared values (mostly above 0.9) over free rollouts of several minutes. If correct, this would let control and reinforcement-learning pipelines be developed and tested against a drop-in surrogate rather than against expensive, scarce physical hardware.

What carries the argument

The central object is the input-output operator formulation (Eq. 2), where the excavator is treated as a black-box map from recent joint states and control commands to future states, approximated by an LSTM with a 30-step history window. The supporting mechanism is the consistency-aware adaptive Kalman filter (Algorithm 1) that fuses noisy angle measurements with integrated angular velocity, adaptively trusting either the angle observation or the velocity integration based on jump detection and window-consistency checks. The filter's output angle is numerically differentiated (Eq. 4) to produce the angular-velocity training target, enforcing kinematic consistency between the angle and veloci

What would settle it

Run the same autoregressive surrogate in a closed loop with a controller that uses the surrogate's angle predictions to plan or react, and compare task-level outcomes (e.g., digging cycle completion, trajectory tracking error) against the physical excavator under identical conditions. If the surrogate's predictions are used to close a feedback loop and the physical machine diverges or becomes unstable, the claimed drop-in functional equivalence would fail. Alternatively, compare the Kalman-fused angle trajectory against independent high-precision external motion capture on the bucket joint, wh

Watch

Extended reading notes

Core claim

The central claim is behavioral equivalence under closed-loop autoregressive evaluation: given the same normalized joint-level control commands, a two-layer stacked LSTM (hidden size 128, input window of 30 steps) that outputs joint angular velocities can be rolled out freely to reproduce the excavator's closed-loop response. In simulation, angle-trajectory R-squared exceeds 0.94 over 90 s. On the real machine, the surrogate achieves angle-trajectory R-squared above 0.92 in no-load conditions and mostly above 0.73 in compound digging operations over 295-383 s, while matching both short-term velocities and long-horizon integrated angles. The paper further claims that its consistency-aware Kal

Load-bearing premise

The whole evaluation rests on treating the angular velocity computed as the numerical derivative of the Kalman-fused angle as the ground truth; if that fused signal is systematically biased or over-smoothed relative to the excavator's actual physical motion, the reported 'fidelity' measures agreement with the filter's manufactured signal, not with the real machine.

Editorial extensions

If this is right

If the surrogate truly reproduces closed-loop behavior, control policies and reinforcement-learning training loops can be run against the surrogate instead of the physical excavator, greatly increasing iteration speed and reducing hardware wear and risk.
The same operator-learning recipe could apply to other hydraulically actuated heavy machinery with limited sensor instrumentation, provided the consistency-enforcement pre-processing is adapted to their sensors.
Long-horizon autoregressive stability, which the paper achieves through a bias-penalized multi-step loss, becomes a practical way to assess whether a learned dynamics model is trustworthy for closed-loop use.
A single learned model could serve both simulation and reality, potentially absorbing simulator-specific corrections and real-world effects into one data-driven operator, simplifying transfer.
The approach sidesteps explicit system identification of hydraulic parameters, meaning it could work even when internal dynamics are unknown or poorly parameterized.

Reading between the lines

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

The strongest version of the claim—that the surrogate is a drop-in replacement for the physical machine—is only as strong as the evaluation protocol: the tests use fixed control inputs and recorded trajectories, not closed-loop control that reacts to the surrogate's own outputs, so true policy-in-the-loop equivalence is not yet demonstrated.
Because the training target is the numerically differentiated Kalman-filtered angle, any smoothing or bias introduced by fusion (most visibly on the bucket joint, with fusion RMSE 0.1093 rad versus under 0.012 rad elsewhere) becomes part of the ground truth; the reported fidelity partly measures agreement with a filtered signal rather than with raw physical measurements.
A natural next experiment would be to run a learning-based controller trained on the surrogate directly on the physical machine and measure task success, which would test whether the surrogate preserves the input-output behavior required for control, not just open-loop reproduction.
The paper's framework suggests that the separator between simulation and reality is not the physics engine but the data consistency pipeline; if so, the MuJoCo stage may be replaceable by any reasonably realistic simulator, or even by recorded real data alone.

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.

Desk Editor's Note

read the letter

The genuinely new thing here is the system-level result: a 4-DoF full-size excavator surrogate, trained on Kalman-consistency-cleaned data from a real SY750H, evaluated by long-horizon free rollout in both MuJoCo and the physical machine. That specific combination is not in the cited literature, and the held-out autoregressive evaluation is the right protocol. Angle R2 above 0.9 over 300+ s on most joints is non-trivial and probably useful to anyone doing RL or controls development for heavy machinery without constant access to a machine.

What the paper does well: it treats the excavator as an input–output operator rather than pretending to model hydraulics; the staged sim-to-real validation is sensible; and the consistency-aware Kalman preprocessing is a practical answer to the real problem that raw inclinometers and rate gyros contradict each other. The multi-step rollout training with the bias-penalty term is also reasonable.

Soft spots, in proportion: first, the stress-test note is right that the evaluation target is the filter's own output. Eq. (4) differentiates the fused angle, and Section IV explicitly says the goal is not estimation accuracy but consistency. So the R2 numbers measure how well the LSTM reproduces the Kalman-filtered signal, not how well it matches raw sensor truth. The bucket fusion RMSE of 0.1093 rad is an order of magnitude larger than the other joints, which is a red flag that the filter can distort the signal. This is not fatal—if the surrogate is meant to replace the filtered pipeline, it may still be the right object to learn—but it means the abstract's 'drop-in surrogate for physical systems' is too strong. The paper should either validate against an independent ground truth or qualify the claim.

Second, there are no baselines. No physics model, no one-step LSTM, no without-Kalman variant. That makes it hard to know what the consistency filter and the bias term actually buy you. Third, the compound-condition hyperparameters (H=10, lambda_bias=0.3) were chosen after observing drift on that very condition, and there are no error bars or multiple seeds, so the reported numbers are point estimates. None of these are fatal, but they are all addressable.

The core approach holds up: the surrogate tracks the system well under free rollout, and the paper is honest about where it had to add tricks to stop drift. I would send this to peer review, because the problem is important, the empirical scope is rare, and a good referee can push the authors to tighten the claims and add the missing comparisons.

For a reading group: maybe, if someone in the group works on sim-to-real for heavy machinery. I probably would not cite it in my own next paper, but only because it is one step outside my immediate area.

Referee Report

3 major / 5 minor

Summary. The paper proposes a system-level surrogate for a four-degree-of-freedom hydraulic excavator using a two-layer LSTM. The excavator is treated as an input–output operator mapping a window of joint states and normalized control commands to joint angular velocities. Training uses a multi-step rollout objective with an optional bias-penalty term. The framework is first validated on a MuJoCo simulation and then on a real SY750H excavator, where an adaptive Kalman filter with consistency detection preprocesses noisy angle/velocity measurements. The surrogate is evaluated autoregressively under held-out control commands over horizons of 90 s (simulation), 382.85 s (no-load real), and 295 s (compound real), reporting RMSE, MAE, and R² for both angular velocity and integrated angle trajectories. The paper claims the learned model can serve as a drop-in surrogate for both the simulator and the physical machine in closed-loop learning and control pipelines.

Significance. If the central claim is correct, the paper addresses a real bottleneck: limited access to heavy hydraulic machines for iterative development of autonomous excavation algorithms. Treating the excavator as a black-box operator and learning an LSTM surrogate from conveniently measured joint states is an appealing and potentially practical approach. The paper has notable methodological strengths: the evaluation is a genuine held-out autoregressive rollout under unseen control commands, the horizons are long (hundreds of seconds), and per-joint metrics are reported for both velocity and integrated angle trajectories. The simulation stage provides a clean sanity check of the operator-learning paradigm. However, the significance is contingent on whether the reported fidelity reflects agreement with the physical machine or only with the paper's own Kalman-filtered signal. As written, the load-bearing 'drop-in surrogate for physical systems' claim is not supported by the evidence presented.

major comments (3)

[§IV-B, Eq. (4), §V-C, Table III] The evaluation ground truth for both training and closed-loop evaluation is the Kalman-filtered angle θ̂ and its numerical derivative ω̂_t=(θ̂_t−θ̂_{t−1})/T_s. Section IV explicitly states that the filter's objective is 'not to improve estimation accuracy per se, but to enforce physically consistent training data.' The reported bucket-joint fusion RMSE of 0.1093 rad (~6.3°) versus ~0.004 rad for the other joints (Fig. 3) shows that the filter can deviate substantially from raw measurements. Because Eq. (4) makes the velocity and integrated-angle targets mutually consistent by construction, the R² values in Table III (e.g., angle R²>0.92) primarily measure how well the LSTM reproduces the filter output, not how well it reproduces the physical machine. The Abstract and Conclusion claim that the surrogate can replace the physical system; this requires validation against raw sensor data or i
[§III-B, §V-D, Table I] The paper claims that the bias-penalized multi-step rollout training improves long-horizon stability and that consistency-aware state estimation 'significantly improves the stability and reliability of surrogate learning.' However, no ablation or baseline comparisons are provided. Table I shows that Stage I and the no-load real experiment use H=1, λ_bias=0, while the compound experiment uses H=10, λ_bias=0.3, but there is no comparison of H=1 vs. H=10, no λ_bias sweep, no variant trained without the Kalman preprocessing, and no physics-based or one-step-trained baseline. These ablations are necessary to support the claimed contributions and to rule out the possibility that the reported performance comes mainly from the filter or from the particular training schedule rather than from the proposed method.
[§V-A–V-C, Tables II–III] All quantitative results are point estimates from a single autoregressive rollout per condition. No error bars, confidence intervals, multiple seeds, or repeated trials are reported. Given the paper's strong claim that the surrogate is a functional replacement for a physical excavator, the absence of any uncertainty quantification is a serious gap. At a minimum, the authors should report statistics over multiple random initializations of the LSTM and, ideally, over multiple held-out test trajectories. This is especially important for the velocity metrics, where some joints already show modest R² values (e.g., boom velocity R²=0.77 in simulation, bucket velocity R²=0.58 in no-load).

minor comments (5)

[§II-B, Eq. (2)] Equation (2) writes the surrogate as producing \tilde{x}_{t+1}, but the model output is defined as joint angular velocity \tilde{y}_t, and the network input includes states and controls. The notation should be made consistent.
[Table III] The table entries are written as 'No-load / Compound' without column headers; this is easy to misread. Consider splitting into two tables or adding explicit subcolumns.
[Fig. 4] The vertical scales for bucket velocity differ markedly between the no-load panel (range about ±5 rad/s) and the compound panel (range about −1 to 0 rad/s), making visual comparison of fidelity difficult. Use consistent scales or clearly annotate the difference.
[§III-A, §V] The term 'simulation-to-real' is potentially misleading: the MuJoCo model is used only to validate the methodology, and the real-world model is retrained on real data. No parameters learned in MuJoCo are transferred. Clarify that the paper provides a staged methodology validation, not a true transfer of a trained model.
[§IV-A, Algorithm 1] The thresholds τ_jump, τ_cons, the noise weights R_trust/R_untrust, and the process noise Q are not given numerically and no sensitivity study is provided. At least list the values used and mention how they were selected.

Circularity Check

2 steps flagged · score 3.0 of 10

Evaluation ground truth is the filter's own output; one-step velocity target is a finite difference of the filtered input angle.

self definitional [Eq. (4), Section IV-B; evaluation configuration in Section V-C.1]
"After fusion, the angle signal is smooth and free of discontinuities, and the angular velocity is recovered via numerical differentiation: ˆωt = (ˆθt − ˆθt−1)/Ts. (4) The fused states (ˆθt, ˆωt) replace the raw sensor readings in the LSTM input vector xt defined in Eq. (2), ensuring that the surrogate model is trained on kinematically consistent data."

The target angular velocity is defined as the derivative of the same filtered angle that enters the LSTM input, and Section IV says the filter's objective is 'not to improve estimation accuracy per se, but to enforce physically consistent training data.' Section V-C.1 then uses the processed data as ground truth for training and evaluation, reporting filter-to-raw RMSE up to 0.1093 rad for the bucket. Thus Table III and Fig. 4 measure how well the LSTM reproduces Algorithm 1's manufactured kinematic signal, not how well it reproduces raw physical sensors. The conclusion that the surrogate is a 'drop-in' replacement for the physical machine therefore rests on the unvalidated premise that the filter output equals the true machine state.
fitted input called prediction [Section V.C.3 (compound operation) and Table I / Table III]
"Initial experiments reveal that standard training leads to trajectory drift, particularly in the swing joint, due to systematic bias in angular velocity prediction. To address this issue, the bias-penalization term in the loss function (Section III-B) is activated (H=10, λbias = 0.3), effectively suppressing long-horizon drift."

The compound-condition hyperparameters (H=10 and λbias=0.3) are selected after observing the exact failure mode that the compound-condition evaluation then reports as fixed. Table III's 'Compound' column is generated by this corrected model on the same operating condition, without a separate calibration/holdout split being stated. The improvement is therefore partly a consequence of the loss being tuned to the evaluation scenario, although the free-rollout evaluation itself remains a genuine prediction.

full rationale

The central LSTM surrogate is not trivially equivalent to its own training targets: Stage I is checked against MuJoCo ground truth on unseen step commands, and Stage II uses held-out autoregressive rollouts with no external correction, so the model is not simply re-reporting fitted outputs. The main circularity-related weakness is that, on the physical machine, the 'ground truth' is produced by the authors' own adaptive Kalman filter, whose stated purpose is consistency enforcement rather than faithful estimation; because the velocity target (Eq. 4) is a finite difference of the input angle, the evaluation partly verifies self-consistency with the filter pipeline. A second, milder leakage is the compound-condition tuning of H and λbias after observing drift on that same condition. These issues raise the score above the 0–2 band, but they do not make the whole derivation circular: the long-horizon free-rollout behavior still requires the LSTM to generalize beyond the training signal.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

No new physical entities, forces, or conserved quantities are introduced; the 'consistency flag' and 'digital surrogate' are methodological constructs, not invented entities. The input cost sits in the sensor-fusion pipeline that manufactures the training/evaluation target (unnamed thresholds and noise weights), the scenario-dependent loss weights tuned after observing the evaluated condition, and the unlisted MuJoCo parameters. Together these mean the reader cannot tell how much of the reported fidelity comes from the LSTM itself versus from preprocessing and hyperparameter choices.

free parameters (5)

lambda_bias = 0.3
Bias-penalty weight in the loss (Section III-B); activated after 'initial experiments' showed trajectory drift on the compound condition, then evaluated on that same condition. No ablation or separate validation split is reported.
rollout horizon H = 1 (sim, no-load); 10 (compound)
Scenario-dependent training horizon; H=10 is coupled with lambda_bias=0.3 to suppress drift on the compound condition. Changed after observing drift, i.e., tuned on the evaluated regime.
Kalman thresholds and noise weights = tau_jump, tau_cons, R_trust, R_untrust, Q_theta, Q_omega (values not reported)
Hand-set inputs to Algorithm 1 that determine when angle measurements are trusted, and therefore shape the fused signal used as both training target and evaluation ground truth. No sensitivity analysis.
MuJoCo actuation latency = 50 ms
Fixed delay introduced in Stage I, 'consistent with reported delay ranges in hydraulic excavators' [12]; chosen from the literature, not measured on this machine.
LSTM hyperparameters = 2 layers, hidden 128, W=30, lr=1e-3
Hand-chosen; no ablation. W=30 implicitly assumes the input-output operator's memory is about 1.5 s at 20 Hz, and W=30 is also used for the real system at 10 Hz (3 s).

assumptions (4)

domain assumption The MuJoCo rigid-body model, built from unlisted geometric and inertial parameters plus a fixed 50 ms delay, adequately represents excavator behavior for validating the operator-learning paradigm.
Invoked in Section III-A1 as Stage I validation. The delay is chosen, not measured; the model parameters are not disclosed, so the 'ideal condition' baseline is unverifiable.
domain assumption Kinematic consistency between angle and integrated angular velocity is a valid physical constraint on the real excavator's states; violations are sensor artifacts.
Section IV: the filter's stated purpose is to enforce consistency rather than improve estimation accuracy. If real hydraulic dynamics or mounting conditions legitimately violate theta = integral(omega), the fused output is a manufactured target, not ground truth.
standard math A constant-velocity transition model (A = [[1,T_s],[0,1]] with process noise Q = diag(Q_theta, Q_omega)) adequately describes joint motion between samples at 10–20 Hz.
Algorithm 1, Section IV-A. Standard Kalman smoothing assumption; its adequacy for fast, load-varying bucket motion is untested.
domain assumption An LSTM with a 30-step sliding window and velocity output can represent the excavator's input-output operator, including hydraulic delay and nonlinearities.
Section III-B, Eq. (2). No analysis links W to the system's memory or delay; only empirical results on this machine support it.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Learning a System-Level Surrogate for Hydraulic Excavators: A Simulation-to-Real LSTM Approach." pith.science (2026). https://pith.science/paper/RAVFQSBU

@misc{pith2026260715656,
  author       = {Pith},
  title        = {Pith review of: Learning a System-Level Surrogate for Hydraulic Excavators: A Simulation-to-Real LSTM Approach},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RAVFQSBU}},
  note         = {Machine review of arXiv:2607.15656}
}

read the original abstract

Developing autonomous hydraulic excavators is constrained by limited access to physical machines and the high cost of real-world experimentation. This paper proposes a simulation-to-real framework for learning a system-level digital surrogate using Long Short-Term Memory (LSTM) networks. Instead of modeling internal dynamics, the excavator is treated as an input-output operator, and the surrogate is trained to reproduce its closed-loop behavior under identical control inputs. The approach is first validated in a MuJoCo simulation environment and then transferred to a real excavator. To address measurement inconsistencies in real-world data, a consistency-aware state estimation method based on adaptive Kalman filtering is introduced. Experimental results demonstrate that the learned surrogate achieves high fidelity in both angular velocity and long-horizon trajectory reproduction under closed-loop autoregressive evaluation. These results confirm that the proposed model can serve as a drop-in surrogate for both simulation and physical systems, enabling scalable and efficient development of excavation automation algorithms.

Figures

Figures reproduced from arXiv: 2607.15656 by the authors.

**Figure 2.** Closed-loop validation in simulation: comparison between the LSTM [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 4.** Closed-loop validation on the physical excavator. Comparison between LSTM free-evolution trajectories and real system responses under identical [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

17 extracted references · 2 linked inside Pith

[1]

An autonomous excavator system for material loading tasks,

L. Zhanget al., “An autonomous excavator system for material loading tasks,”Science Robotics, vol. 6, no. 55, p. eabc3164, Jun. 2021, doi: 10.1126/scirobotics.abc3164

doi:10.1126/scirobotics.abc3164 2021
[2]

A general approach for the automation of hydraulic excavator arms using reinforcement learning,

P. Egli and M. Hutter, “A general approach for the automation of hydraulic excavator arms using reinforcement learning,”IEEE Robot. Autom. Lett., vol. 7, no. 2, pp. 5679–5686, Apr. 2022

2022
[3]

Data-driven modeling and control for the automation of industrial machinery with limited instrumentation,

N. Hoffmann, “Data-driven modeling and control for the automation of industrial machinery with limited instrumentation,” Ph.D. dissertation, Institut Polytechnique de Paris, 2024

2024
[4]

Precision motion control of robotized industrial hydraulic excavators via data- driven model inversion,

M. Lee, H. Choi, C. Kim, J. Moon, D. Kim, and D. Lee, “Precision motion control of robotized industrial hydraulic excavators via data- driven model inversion,”IEEE Robot. Autom. Lett., vol. 7, no. 2, pp. 1912–1919, Apr. 2022

1912
[5]

Identification and control of dynamical systems using neural networks,

K. S. Narendra and K. Parthasarathy, “Identification and control of dynamical systems using neural networks,”IEEE Trans. Neural Netw., vol. 1, no. 1, pp. 4–27, Mar. 1990

1990
[6]

Robot model identification and learning: A modern perspective,

T. Lee, J. Kwon, P. M. Wensing, and F. C. Park, “Robot model identification and learning: A modern perspective,”Annu. Rev. Control Robot. Auton. Syst., vol. 7, no. 1, pp. 311–334, Jul. 2024

2024
[7]

A data-driven approach for approximating non-linear dynamic systems using LSTM networks,

L. Heindel, P. Hantschke, and M. K ¨astner, “A data-driven approach for approximating non-linear dynamic systems using LSTM networks,” Procedia Struct. Integr., vol. 38, pp. 159–167, 2022

2022
[8]

LSTM-based adaptive robust nonlinear controller design of a single-axis hydraulic shaking table,

J. Wen, C. Zhao, and Z. Shi, “LSTM-based adaptive robust nonlinear controller design of a single-axis hydraulic shaking table,”IET Control Theory Appl., vol. 17, no. 7, pp. 825–836, Apr. 2023

2023

Show all 17 references

[9]

Data-driven multi-step nonlinear model predictive control for industrial heavy load hydraulic robot,

D. Ma and B. Zhou, “Data-driven multi-step nonlinear model predictive control for industrial heavy load hydraulic robot,” Nov. 2024, arXiv: 2411.13859

2024 arXiv
[10]

Modeling weakly- instrumented excavator arm dynamics with stacked-input LSTM,

N. Hoffmann, M. Cohen, M. Preda, and T. Zaharia, “Modeling weakly- instrumented excavator arm dynamics with stacked-input LSTM,” in Proc. 12th Int. Conf. Robot Intell. Technol. Appl. (RiTA), 2024

2024
[11]

Examining the simulation-to-reality gap of a wheel loader digging in deformable terrain,

K. Aoshima and M. Servin, “Examining the simulation-to-reality gap of a wheel loader digging in deformable terrain,”Multibody Syst. Dyn., vol. 64, no. 1, pp. 121–148, May 2025

2025
[12]

A position controller for hydraulic excavators with deadtime and regenerative pipelines,

Y . Yamamoto, J. Qiu, T. Doi, T. Nanjo, K. Yamashita, and R. Kikuuwe, “A position controller for hydraulic excavators with deadtime and regenerative pipelines,”IEEE Trans. Autom. Sci. Eng., vol. 22, pp. 855– 871, 2025

2025
[13]

Data-driven identification of nonlinear dynamical systems with LSTM autoencoders and normaliz- ing flows,

A. Rostamijavanani, S. Li, and Y . Yang, “Data-driven identification of nonlinear dynamical systems with LSTM autoencoders and normaliz- ing flows,” Mar. 2025, arXiv: 2503.03977

2025 arXiv
[14]

Smoothing and stationarity enforcement framework for deep learning time-series forecasting,

I. E. Livieris, S. Stavroyiannis, L. Iliadis, and P. Pintelas, “Smoothing and stationarity enforcement framework for deep learning time-series forecasting,”Neural Comput. Appl., vol. 33, no. 20, pp. 14021–14035, Oct. 2021

2021
[15]

An enhanced adaptive Kalman filter for multibody model observation,

A. J. Rodr ´ıguez, E. Sanjurjo, and M. ´A. Naya, “An enhanced adaptive Kalman filter for multibody model observation,”Sensors, vol. 25, no. 7, p. 2218, Apr. 2025

2025
[16]

Mathematical modelling and virtual decomposition control of heavy-duty parallel−serial hydraulic manip- ulators,

G. R. Petrovi ´c and J. Mattila, “Mathematical modelling and virtual decomposition control of heavy-duty parallel−serial hydraulic manip- ulators,” Mechanism and Machine Theory, vol. 170, p. 104680, Apr. 2022

2022
[17]

MuJoCo: A physics engine for model-based control,

E. Todorov, T. Erez, and Y . Tassa, “MuJoCo: A physics engine for model-based control,” in2012 IEEE/RSJ Int. Conf. Intell. Robots Syst. (IROS), Vilamoura-Algarve, Portugal, Oct. 2012, pp. 5026–5033

2012

Pith tools

Reviewed August 1, 2026 · model on record in the stance chip above.

[1] [1]

An autonomous excavator system for material loading tasks,

L. Zhanget al., “An autonomous excavator system for material loading tasks,”Science Robotics, vol. 6, no. 55, p. eabc3164, Jun. 2021, doi: 10.1126/scirobotics.abc3164

doi:10.1126/scirobotics.abc3164 2021

[2] [2]

A general approach for the automation of hydraulic excavator arms using reinforcement learning,

P. Egli and M. Hutter, “A general approach for the automation of hydraulic excavator arms using reinforcement learning,”IEEE Robot. Autom. Lett., vol. 7, no. 2, pp. 5679–5686, Apr. 2022

2022

[3] [3]

Data-driven modeling and control for the automation of industrial machinery with limited instrumentation,

N. Hoffmann, “Data-driven modeling and control for the automation of industrial machinery with limited instrumentation,” Ph.D. dissertation, Institut Polytechnique de Paris, 2024

2024

[4] [4]

Precision motion control of robotized industrial hydraulic excavators via data- driven model inversion,

M. Lee, H. Choi, C. Kim, J. Moon, D. Kim, and D. Lee, “Precision motion control of robotized industrial hydraulic excavators via data- driven model inversion,”IEEE Robot. Autom. Lett., vol. 7, no. 2, pp. 1912–1919, Apr. 2022

1912

[5] [5]

Identification and control of dynamical systems using neural networks,

K. S. Narendra and K. Parthasarathy, “Identification and control of dynamical systems using neural networks,”IEEE Trans. Neural Netw., vol. 1, no. 1, pp. 4–27, Mar. 1990

1990

[6] [6]

Robot model identification and learning: A modern perspective,

T. Lee, J. Kwon, P. M. Wensing, and F. C. Park, “Robot model identification and learning: A modern perspective,”Annu. Rev. Control Robot. Auton. Syst., vol. 7, no. 1, pp. 311–334, Jul. 2024

2024

[7] [7]

A data-driven approach for approximating non-linear dynamic systems using LSTM networks,

L. Heindel, P. Hantschke, and M. K ¨astner, “A data-driven approach for approximating non-linear dynamic systems using LSTM networks,” Procedia Struct. Integr., vol. 38, pp. 159–167, 2022

2022

[8] [8]

LSTM-based adaptive robust nonlinear controller design of a single-axis hydraulic shaking table,

J. Wen, C. Zhao, and Z. Shi, “LSTM-based adaptive robust nonlinear controller design of a single-axis hydraulic shaking table,”IET Control Theory Appl., vol. 17, no. 7, pp. 825–836, Apr. 2023

2023