Meta-Learning Online Dynamics Model Adaptation in Off-Road Autonomous Driving

Bogdan Vlahov; David Fridovich-Keil; Erica Tevere; Evangelos Theodorou; Jacob Levy; Jason Gibson; Patrick Spieler

REVIEW 3 major objections 6 minor 1 cited by

Meta-Learning Online Dynamics Model Adaptation in Off-Road Autonomous Driving

T0 review · 3 major / 6 minor · reviewed 2026-08-16 · 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 meta-learning the directions in which a Kalman filter can adapt a vehicle dynamics model lets an off-road car correct its model in real time, lowering 5-second prediction error from 4.88 m to 3.10 m and sharply…

desk verdict Useful combination of meta-learning and Kalman filter adaptation for off-road dynamics, with real-vehicle evidence; the 'unseen terrain' claim outruns the experiment. read the letter →

arxiv 2504.16923 v1 pith:OFTKUXKV submitted 2025-04-23 cs.RO cs.LGcs.SYeess.SY

Jacob Levy , Jason Gibson , Bogdan Vlahov , Erica Tevere , Evangelos Theodorou , David Fridovich-Keil , Patrick Spieler This is my paper · ORCID

classification cs.ROcs.LGcs.SYeess.SY

keywords meta-learningonlineadaptationKalmanfilteroff-roadautonomousdrivingmodelpredictivepathintegralcontrolvehicledynamicssafety-criticalterrain

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 tackles a concrete failure mode: a dynamics model trained offline drives well until the terrain changes, after which a model-based controller acts on wrong predictions. It proposes to split the problem in two, with offline meta-learning choosing a small set of 'directions' in the model-parameter space and an online Kalman filter moving the parameters along those directions using live state measurements. The claim is that this split makes real-time adaptation fast enough for high-speed off-road driving while staying accurate. The supporting evidence is a full-scale vehicle experiment in which the adapted model's 5-second endpoint prediction error falls to 3.10 m from 4.88 m and the time spent beyond track and rollover safety limits drops sharply.

What carries the argument

The central object is the tensor $W \in \mathbb{R}^{n_w \times n_{\mathrm{out}} \times n_{\mathrm{in}}}$ formed by stacking the last-layer weight matrices of the feedforward part of the learned dynamics model; the paper treats these as adaptation basis functions. The dynamics are linear in the adaptable parameters $\theta = [\theta_w, \theta_b]$, which is what makes a Kalman filter a natural online estimator, and a recursive multi-step Jacobian $H_{t+h} = \partial \hat{x}_{t+h}/\partial \theta_t$ lets the filter see how parameter changes affect predictions several steps ahead. Offline, gradient-based meta-learning backpropagates through the whole adaptation procedure to tune $W$, the initial covariance $P_s$, the noise matrices $Q$ and $R$, and the speed-scaling constant $\varepsilon$; online, all the filter does is update the coefficients along those learned directions.

What would settle it

Run the trained model on a terrain type whose vehicle response changes in a direction the learned corrections cannot represent, such as deep snow, and check whether 5-second endpoint prediction error and rollover-limit violations still improve over no adaptation; if they do not, the offline coverage assumption is the point of failure. A cheaper offline check is to compare the size of the part of the measured model mismatch that lies outside the learned correction directions with the size of the whole mismatch.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central discovery is that the adaptation directions matter as much as the adaptation itself. The last-layer weight ensemble $W$ of the learned dynamics model is not just a set of parameters; it is the subspace the Kalman filter is allowed to move in, and meta-learning $W$ together with the Kalman filter noise matrices $Q$ and $R$ determines where and how fast adaptation happens. With this meta-learned setup, the online system reduces endpoint prediction error and, in closed loop with the model predictive path integral control (MPPI) controller, the vehicle completes a 3-lap course faster while crossing track boundaries and rollover thresholds far less often than the no-adaptation baseline. The simulated experiments add the same conclusion in a real-to-sim transfer setting, where the meta-learned configuration generally beats both a non-meta-learned Kalman filter adaptation and a sliding-window least-squares baseline on prediction error and safety metrics.

Load-bearing premise

The load-bearing premise is that the directions of model correction learned offline cover every way the vehicle's behavior can change on new terrain; if some new surface changes the vehicle in a direction the learned corrections do not include, the online filter cannot represent the needed fix and the claimed safety gains would not follow.

Editorial extensions

If this is right

In real vehicle runs, the adapted model cuts the 5-second endpoint prediction error from 4.88 m to 3.10 m while completion time drops from 154.6 s to 130.9 s and average speed rises from 5.06 m/s to 5.84 m/s.
The adapted vehicle crosses track boundaries and rollover limits less often and spends far less time beyond them, which is the safety payoff of accurate rollouts inside MPPI.
In the simulated real-to-sim gap, meta-learned adaptation generally achieves the lowest prediction errors and best or tied rollover-safety metrics across four procedurally generated maps.
Because the method only requires dynamics linear in the adaptable parameters, it transfers to any model-based controller whose rollouts depend on such a model, not only to this off-road vehicle.

Reading between the lines

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

A testable extension is to monitor the projection of the online prediction residual onto the span of $W$; when that projection is small but the residual is large, the learned basis is missing a needed direction and the algorithm could flag that new terrain for offline retraining.
The random-walk noise model for $\theta$ implies the adaptation rate is tuned for a single timescale; terrain that changes faster than the Kalman filter's learned time constant would likely need context-dependent $Q$, something the paper does not explore.
The same 'learn the adaptation subspace, then filter along it' pattern could be applied to other platforms with learned residual dynamics, such as legged robots or rotorcraft, whenever the residual model is linear in the adapted parameters.

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

Good paper to know about if you work on adaptive model-based control for mobile robots. The new bit is a specific architecture: they meta-learn an ensemble of last-layer weight matrices plus Kalman filter covariances and a speed-scaling term, and online only adapt the weights in that learned subspace. That combination on a full-scale RZR is genuinely new in this space, and the reported effect is large: prediction error drops from 4.88 to 3.10 m, safety violations drop sharply. The real2sim transfer experiment is a nice touch because it tests adaptation to a deliberately introduced model mismatch rather than just more of the same terrain. They also compare against a sliding-window least-squares baseline and an adaptation-without-meta-learning baseline, which is the right experimental skeleton.

The soft spots are real but mostly addressable. The real-vehicle comparison is four runs per configuration, no significance testing; with that sample, I'd treat the point estimates as encouraging rather than definitive. The larger issue is the mismatch between the abstract's 'unseen environments' claim and the actual evidence. The adaptation subspace W is meta-learned on 60% Mojave, 30% Halter Ranch, 10% coastal sage, 5% coastal dunes. Remark 5 says the Kalman filter can only adjust parameters in the span of W. If the real test course is from one of those same sites, then Table II only shows adaptation to mismatch directions seen in training. The paper never states the test terrain is outside the offline dataset. The real2sim experiment introduces a global bicycle-model mismatch, which is not the same as a new terrain's mismatch. So the core generalization claim—works on unseen terrain—is plausible but unproven; what they've shown is that meta-learned adaptation works on their dataset's terrain distribution.

There's also a concrete technical issue: in Algorithm 1, line 10, the covariance update is written with a plus sign before the K C H term. Standard Kalman covariance update is \bar P - K C H \bar P. If that plus sign is actually in the code, the filter isn't a Kalman filter. More likely it's a typo, but with no code or data release there's no way to check. Minor: the real2sim baseline is sometimes faster, which they attribute to unsafe behavior, and that trade-off is plausible but not deeply analyzed.

The limitations section is candid about excitation and training cost, which I appreciate. Overall, this is a serious engineering contribution with a clearly explained method and mostly honest reporting. It deserves a proper peer review; I would not desk-reject it. The reviewers should press on the coverage assumption and ask for either a held-out terrain test or a relaxation of the claim. I'd want the sign error clarified before publication. For me personally, I'd cite it as an example of meta-learned KF adaptation for ground vehicles, but I wouldn't cite it as evidence for generalization across unseen terrain.

Referee Report

3 major / 6 minor

Summary. The paper proposes a meta-learning framework for online adaptation of a vehicle dynamics model used in model-based control for high-speed off-road driving. Offline meta-learning optimizes an ensemble of last-layer weight matrices W (the adaptation basis), Kalman filter parameters Q, R, P_s, and epsilon, while online a Kalman filter updates a low-dimensional parameter theta_w that weights this ensemble plus a bias term. The adapted model is used inside MPPI for real-time planning. Experiments on a full-scale Polaris RZR (4 runs per configuration on a figure-8 course) and in a bicycle-model simulator with procedurally generated maps compare the meta-adaptive method against a no-adaptation baseline and, in simulation, against non-meta-learned adaptation baselines. The reported results show lower prediction error and fewer safety violations for the meta-adaptive method, supporting the main direction of the paper.

Significance. If the central claim is accepted, this is a practically valuable contribution: it demonstrates a working, real-time online adaptation scheme that improves both prediction accuracy and closed-loop safety in a full-scale off-road vehicle, and it shows a clear benefit of meta-learning the adaptation structure rather than hand-designing it. The paper is honest about several limitations, including the need for sufficiently exciting inputs and user-defined initial Kalman parameters. The real-world validation on a full-scale vehicle is a notable strength, as are the multi-step Jacobian formulation and the comparison against two non-meta-learned adaptation baselines in simulation. The claim of adaptation to truly unseen terrain, however, is not yet established by the evidence presented.

major comments (3)

[Section V-A2, Table II] The text states that the adapted model achieves 'significantly lower prediction error' and 'significantly less time in unsafe states', but no statistical significance test is reported for the real-world results. With only n=4 runs per configuration, the statement is unsupported; for example, the # times crossed track values are 8.0 ± 1.8 versus 3.3 ± 2.1, where the standard deviations are large relative to the sample size. Please report uncertainty (e.g., bootstrapped confidence intervals or a paired test) analogous to the simulation experiments, or temper the language to 'lower mean values'.
[Section IV-A and Remark 5, with Section V-A1] The manuscript's abstract and introduction claim adaptation to 'unseen' terrain, but the real-world test course is not stated to be outside the offline training distribution, and the design of the adaptation subspace makes this claim load-bearing. Equation (7) and Remark 5 show that online adaptation can only adjust theta_w within the span of the meta-learned ensemble W (plus a constant bias). No coverage or completeness argument is provided for why W should span mismatch directions on a genuinely new terrain. The simulated real2sim experiments introduce a global bicycle-model mismatch, not a terrain-specific mismatch, so they do not test generalization to unseen terrain. Please either (a) explicitly state whether the real-world test course is in the training distribution and, if so, revise the claims accordingly, or (b) provide evidence of W's coverage, for example by testing on a site not used in training or by analyzing the learned W and the mismatch directions observed on the test course.
[Section IV-B, Assumption 3 and Eq. (8)] Assumption 3 states that changes in the learned model output with respect to the state are negligible, i.e., ∂ζ/∂x ≈ 0. This assumption is used in the multi-step Jacobian recursion (8), which directly determines the Kalman gain and the parameter update. Since the learned model ζ includes an LSTM and an FNN whose inputs η_t contain the state x_t, the approximation is not obviously valid, especially during aggressive maneuvers. No empirical justification is provided. Please validate this assumption on the training data, or analyze how violations affect the adaptation update, since an incorrect Jacobian could bias the parameter estimates.

minor comments (6)

[Algorithm 2 and Section IV-C] Algorithm 2's optimization loop (line 10) lists ξ ∈ {φ, ψ, P_s, Q, R}, omitting ε, while the text in Section IV-C states that ε is also meta-learned. Please make the algorithm consistent with the text.
[Remark 4] The decay parameter β is introduced in Remark 4 but its value is not specified, nor is it listed as a learned or tuned parameter. Please clarify how β is set.
[Table II] The cost rows for track and rollover are reported without standard deviations, while the other metrics include them. Please report the variability of the cost metrics or justify their omission.
[Section V-A1] The dataset description states 'approximately 1,700,000 trajectories (9.5 hours)'. Given a time step of 0.02 s and a total of 9.5 hours, this implies heavily overlapping trajectories; please clarify the number of unique runs and how overlap is handled during training.
[Section V-A1] The phrase 'with discrete time steps spaced ( 0.02 s) apart' and similar parentheticals elsewhere (e.g., 'τ = 1,000 steps ( 20 s)') contain stray spaces and are stylistically inconsistent. Please clean up the formatting.
[Section III-B, Eq. (6)] The function P_n(·; r_limit) is not precisely defined. Please specify its functional form or provide a reference, since it is central to the rollover cost.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the meta-learning and online adaptation claims are tested against non-meta baselines in both real-world deployment and simulated real2sim transfer.

full rationale

The paper's derivation chain is not circular by construction. Eq. (7) defines the adapted model output in terms of the meta-learned ensemble W and the online-adapted coefficients θw, θb. Algorithm 2 optimizes W, ψ, Ps, Q, R, and ε by minimizing the multi-step prediction loss Li in Eq. (11). The real-world evaluation in Section V-A2 then measures endpoint prediction error on new deployment runs, and the simulated experiments in Section V-B deliberately use a different bicycle-model dynamics, creating a real2sim transfer scenario that is independent of the offline training distribution. The meta-training loss and the evaluation metric are related, both being multi-step prediction errors, but the comparison is against a no-adaptation baseline and a non-meta Sliding LSQ baseline, so the reported gain is not a fitted value simply renamed as a prediction. No equation in the paper reduces to its own input: the online Kalman filter (Algorithm 1) uses live velocity measurements, and the meta-learned parameters are not constructed from the evaluation outcome. The self-citations to [18] and [17] provide the base dynamics and control architecture; they are not used as a uniqueness theorem or to forbid alternative models, and the present experiments provide external validation of the overall system. The paper does not explicitly state that the real-world test course lies outside the offline training sites, which would be a limitation for the broad 'unseen terrain' claim, but that is a generalization and data-splitting concern rather than a circular derivation. Consequently, no specific circular step can be exhibited, and the appropriate finding is no significant circularity.

Assumptions & free parameters 8 free parameters · 6 assumptions · 0 invented entities

The central claim rests on the learned model parameters, the Kalman filter formulation, and the implicit transfer assumption that W learned offline spans the adaptation directions needed in unseen terrain.

free parameters (8)

Ensemble of last-layer weights W (adaptation basis functions) = meta-learned on the 1.7M-trajectory dataset; values not reported
W determines the directions in parameter space the Kalman filter can adapt (Remark 5). Since it is optimized offline to minimize prediction loss, the online adaptation's expressiveness is restricted to this learned subspace.
Kalman process/adaptation noise covariance Q = meta-learned; values not reported
Governs the time constant of adaptation (Remark 6); optimized with gradient descent through Algorithm 1.
Measurement/process noise covariance R = meta-learned; values not reported
Encapsulates dynamics and measurement noise (Remark 6); strongly affects the Kalman gain and adaptation speed.
Initial covariance P_s = meta-learned; values not reported
Initial uncertainty of adaptable parameters at run start; an input to Algorithm 1.
Speed scaling factor epsilon = meta-learned; values not reported
Adjusts update scaling gamma_t in Equation (10) to avoid updating when the vehicle is slow.
Decay beta in Remark 4 = in (0,1), exact value not reported
Introduced ad hoc for training stability; it is a user-set hyperparameter not grounded in data.
Learned model parameters phi and psi = trained on 9.5 hours of driving data; architectures not reported
These define the LSTM/FNN and parametric tire and drag model; the meta-adaptation layer operates on top of them.
Neural network architecture and meta-learning hyperparameters = not reported (nw, nin, nout, NB, alpha, NE, tau=1000, h=10)
These choices set adaptation timescale and capacity. Without them the method cannot be exactly replicated.

assumptions (6)

domain assumption Adaptable parameters evolve as a random walk: theta_{t+h}=theta_t+w_theta_t with w_theta_t ~ N(0,Q).
Assumption 1 in Section IV.B. Standard for Kalman filtering; if the true parameter dynamics are non-stationary or correlated, the filter mean and covariance are misspecified.
domain assumption System dynamics have additive Gaussian noise: x_{t+1}=f(x_t,u_t,y_t;theta_t)+w_x_t with w_x_t ~ N(0,R).
Assumption 2 in Section IV.B. Used to derive the Kalman update; the paper later treats R as including measurement noise as well (Remark 6), but the formal model is process noise.
ad hoc to paper Changes in the learned model output with respect to the state are negligible: partial zeta / partial x is approximately 0.
Assumption 3 in Section IV.B. Introduced to simplify the multi-step Jacobian recursion (8); no empirical justification is provided, and it can bias the Jacobian if the neural residual is state-sensitive.
domain assumption The true value of theta is approximately constant within each training trajectory.
Assumption 4 in Section IV.C.2. Needed so that running Algorithm 1 from t-tau to t identifies a single theta_t; violated when terrain changes within a 20-second trajectory.
domain assumption The meta-learned basis ensemble W generalizes to unseen terrains.
Implicit in Remark 5 and the abstract's claim of operation in unseen environments; no coverage or transfer argument is provided.
domain assumption The vehicle state is fully observable.
Section III.A states the state x_t is fully observable. In practice it comes from FGO localization with noise and delay, which the online filter is meant to handle but the model assumes direct observations.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Meta-Learning Online Dynamics Model Adaptation in Off-Road Autonomous Driving." pith.science (2026). https://pith.science/paper/OFTKUXKV

@misc{pith2026250416923,
  author       = {Pith},
  title        = {Pith review of: Meta-Learning Online Dynamics Model Adaptation in Off-Road Autonomous Driving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OFTKUXKV}},
  note         = {Machine review of arXiv:2504.16923}
}

read the original abstract

High-speed off-road autonomous driving presents unique challenges due to complex, evolving terrain characteristics and the difficulty of accurately modeling terrain-vehicle interactions. While dynamics models used in model-based control can be learned from real-world data, they often struggle to generalize to unseen terrain, making real-time adaptation essential. We propose a novel framework that combines a Kalman filter-based online adaptation scheme with meta-learned parameters to address these challenges. Offline meta-learning optimizes the basis functions along which adaptation occurs, as well as the adaptation parameters, while online adaptation dynamically adjusts the onboard dynamics model in real time for model-based control. We validate our approach through extensive experiments, including real-world testing on a full-scale autonomous off-road vehicle, demonstrating that our method outperforms baseline approaches in prediction accuracy, performance, and safety metrics, particularly in safety-critical scenarios. Our results underscore the effectiveness of meta-learned dynamics model adaptation, advancing the development of reliable autonomous systems capable of navigating diverse and unseen environments. Video is available at: https://youtu.be/cCKHHrDRQEA

Figures

Figures reproduced from arXiv: 2504.16923 by the authors.

**Figure 1.** Trajectories for a single 3-lap run, with insets displaying video stills. The baseline configuration shows erratic trajectories with frequent course boundary and rollover limit violations. In contrast, our adaptation configuration demonstrates more deliberate and compliant trajectories as the car learns the terrain dynamics in real time. integrity. When system dynamics are known, model-based control techniques are w… view at source ↗

**Figure 2.** Meta-learning online dynamics model adaptation. Online, a Kalman filter updates the linear combination of an ensemble of last-layer weights (Algorithm 1). Offline, trajectory segments are used to meta-learn model parameters, the last-layer ensemble, and the Kalman filter parameters (Algorithm 2). learned components to improve predictive accuracy [12, 25, 13, 16, 30]. While these models are effective under nominal co… view at source ↗

**Figure 3.** Forward facing camera stills from the dataset highlighting a diverse range of terrains: A) flat sandy beach with a mixture of packed wet sand and loose dry sand; B) wet dense mud that forms deep ruts; C) dirt trails with low dry grass that weave through dense trees; D) mixed vegetation including dry, dense vehicle-height grass; E) dense overgrown mixed vegetation ranging in crushability; and F) loose gravel, uneven … view at source ↗

Figures from the paper (3 more)

**Figure 4.** Figure 4: Trajectories for all 3-lap real-world runs. The baseline configuration exhibits erratic motion, frequently violating course boundaries and rollover limits. In contrast, our adaptation configuration produces more deliberate and compliant trajectories, as the vehicle lea…

**Figure 5.** Figure 5: Norm of the adapted parameters during one of the real-world runs. Hypothesis 2. In high-speed off-road driving, an inaccurate dynamics model can lead MPPI to generate unsafe trajectories that fail to account for real-world dynamics. By incorporating accurate, environm…

**Figure 6.** Figure 6: Procedurally generated maps for the simulated experiments; horizontal and vertical axes are in meters. configuration on each map. 2) Results: The results of the simulated experiments are summarized in Table III. As expected, prediction errors in simulation are generall…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

Adapting Generalist Vehicle Models for High-Speed MPC Across Terrains
cs.RO 2026-07 conditional novelty 6.0 of 10

History-conditioned fine-tuning with targeted synthetic rollouts from a per-terrain bicycle model roughly halves 6 m/s trajectory tracking error against a fine-tuned AnyCar baseline.

Reference graph

Works this paper leans on

52 extracted references · 42 canonical work pages · cited by 1 Pith paper

[1]

Online system identification and calibration of dynamic models for autonomous ground vehicles

Sina Aghli and Christoffer Heckman. Online system identification and calibration of dynamic models for autonomous ground vehicles. In 2018 IEEE International Conference on Robotics and Automation (ICRA) , pages 4933–4939. IEEE, 2018

work page 2018
[2]

Learning and prediction of slip from visual information

Anelia Angelova, Larry Matthies, Daniel Helmick, and Pietro Perona. Learning and prediction of slip from visual information. Journal of Field Robotics , 24(3):205–231, 2007

work page 2007
[3]

Few-shot semantic learning for robust multi-biome 3d seman- tic mapping in off-road environments

Deegan Atha, Xianmei Lei, Shehryar Khattak, Anna Sabel, Elle Miller, Aurelio Noca, Grace Lim, Jeffrey Edlund, Curtis Padgett, and Patrick Spieler. Few-shot semantic learning for robust multi-biome 3d seman- tic mapping in off-road environments. arXiv preprint arXiv:2411.06632, 2024

arXiv 2024
[4]

Vehicle modeling and state estimation for autonomous driving in terrain

Tabish Badar, Juha Backman, and Arto Visala. Vehicle modeling and state estimation for autonomous driving in terrain. Control Engineering Practice, 152:106046, 2024

work page 2024
[5]

Adaptive Meta-Learning for Identification of Rover-Terrain Dynamics

Somrita Banerjee, James Harrison, P Michael Furlong, and Marco Pavone. Adaptive meta-learning for iden- tification of rover-terrain dynamics. arXiv preprint arXiv:2009.10191, 2020

work page Pith review arXiv 2009
[6]

Autonomous vehicles and intelligent automa- tion: Applications, challenges, and opportunities

Gourav Bathla, Kishor Bhadane, Rahul Kumar Singh, Rajneesh Kumar, Rajanikanth Aluvalu, Rajalakshmi Kr- ishnamurthi, Adarsh Kumar, RN Thakur, and Shakila Basheer. Autonomous vehicles and intelligent automa- tion: Applications, challenges, and opportunities. Mobile Information Systems, 2022(1):7632892, 2022

work page 2022
[7]

Model-based meta-reinforcement learning for flight with suspended payloads

Suneel Belkhale, Rachel Li, Gregory Kahn, Rowan McAllister, Roberto Calandra, and Sergey Levine. Model-based meta-reinforcement learning for flight with suspended payloads. IEEE Robotics and Automation Letters, 6(2):1471–1478, 2021

work page 2021
[8]

Risk-aware off-road navigation via a learned speed distribution map

Xiaoyi Cai, Michael Everett, Jonathan Fink, and Jonathan P How. Risk-aware off-road navigation via a learned speed distribution map. In 2022 IEEE/RSJ Inter- national Conference on Intelligent Robots and Systems (IROS), pages 2931–2937. IEEE, 2022

work page 2022

Show all 52 references

[9]

Probabilistic traversability model for risk-aware motion planning in off-road envi- ronments

Xiaoyi Cai, Michael Everett, Lakshay Sharma, Philip R Osteen, and Jonathan P How. Probabilistic traversability model for risk-aware motion planning in off-road envi- ronments. In 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 11297– 113...

2023
[10]

A note on two problems in connexion with graphs

Edsger W Dijkstra. A note on two problems in connexion with graphs. Numerische mathematik , 1(1):269–271, 1959

1959
[11]

Learning-based on-track system identification for scaled autonomous racing in under a minute

Onur Dikici, Edoardo Ghignone, Cheng Hu, Nicolas Baumann, Lei Xie, Andrea Carron, Michele Magno, and Matteo Corno. Learning-based on-track system identification for scaled autonomous racing in under a minute. arXiv preprint arXiv:2411.17508 , 2024

2024 arXiv
[12]

Neural networks with physics- informed architectures and constraints for dynamical sys- tems modeling

Franck Djeumou, Cyrus Neary, Eric Goubault, Sylvie Putot, and Ufuk Topcu. Neural networks with physics- informed architectures and constraints for dynamical sys- tems modeling. In Learning for Dynamics and Control Conference, pages 263–277. PMLR, 2022

2022
[13]

Autonomous drifting with 3 minutes of data via learned tire models

Franck Djeumou, Jonathan YM Goh, Ufuk Topcu, and Avinash Balachandran. Autonomous drifting with 3 minutes of data via learned tire models. In 2023 IEEE International Conference on Robotics and Automation (ICRA), pages 968–974. IEEE, 2023

2023
[14]

Rose: Robust state estimation via online covariance adaption

Seyed Fakoorian, Kyohei Otsu, Shehryar Khattak, Matteo Palieri, and Ali-akbar Agha-mohammadi. Rose: Robust state estimation via online covariance adaption. In The International Symposium of Robotics Research , pages 452–467. Springer, 2022

2022
[15]

Model- agnostic meta-learning for fast adaptation of deep net- works

Chelsea Finn, Pieter Abbeel, and Sergey Levine. Model- agnostic meta-learning for fast adaptation of deep net- works. In International conference on machine learning , pages 1126–1135. PMLR, 2017

2017
[16]

A multi-step dynamics mod- eling framework for autonomous driving in multiple environments

Jason Gibson, Bogdan Vlahov, David Fan, Patrick Spieler, Daniel Pastor, Ali-akbar Agha-mohammadi, and Evangelos A Theodorou. A multi-step dynamics mod- eling framework for autonomous driving in multiple environments. In 2023 IEEE International Conference on Robotics and Automa...

2023
[17]

Dynam- ics modeling using visual terrain features for high- speed autonomous off-road driving

Jason Gibson, Anoushka Alavilli, Erica Tevere, Evan- gelos A Theodorou, and Patrick Spieler. Dynam- ics modeling using visual terrain features for high- speed autonomous off-road driving. arXiv preprint arXiv:2412.00581, 2024

2024 arXiv
[18]

Multistep belief space dynamics learning for risk-aware control

Jason Gibson, Bogdan Vlahov, Patrick Spieler, and Evan- gelos A Theodorou. Multistep belief space dynamics learning for risk-aware control. arXiv preprint, 2025

2025
[19]

Model predictive control for aggressive driving over uneven terrain

Tyler Han, Alex Liu, Anqi Li, Alex Spitzer, Guanya Shi, and Byron Boots. Model predictive control for aggressive driving over uneven terrain. arXiv preprint arXiv:2311.12284, 2023

2023 arXiv
[20]

Continuous meta-learning without tasks

James Harrison, Apoorva Sharma, Chelsea Finn, and Marco Pavone. Continuous meta-learning without tasks. Advances in neural information processing systems , 33: 17571–17581, 2020

2020
[21]

Meta-learning priors for efficient online bayesian re- gression

James Harrison, Apoorva Sharma, and Marco Pavone. Meta-learning priors for efficient online bayesian re- gression. In Algorithmic Foundations of Robotics XIII: Proceedings of the 13th Workshop on the Algorithmic Foundations of Robotics 13 , pages 318–337. Springer, 2020

2020
[22]

Terrain esti- mation for high-speed rough-terrain autonomous vehicle navigation

Karl D Iagnemma and Steven Dubowsky. Terrain esti- mation for high-speed rough-terrain autonomous vehicle navigation. In Unmanned Ground Vehicle Technology IV, volume 4715, pages 256–266. SPIE, 2002

2002
[23]

Robust adaptive control

Petros A Ioannou and Jing Sun. Robust adaptive control. Courier Corporation, 2012

2012
[24]

Fast nonlinear model predictive planner and control for an unmanned ground vehicle in the presence of disturbances and dynamic obstacles

Subhan Khan and Jose Guivant. Fast nonlinear model predictive planner and control for an unmanned ground vehicle in the presence of disturbances and dynamic obstacles. Scientific Reports, 12(1):12135, 2022

2022
[25]

Physics embedded neural network vehicle model and applications in risk-aware autonomous driving using latent features

Taekyung Kim, Hojin Lee, and Wonsuk Lee. Physics embedded neural network vehicle model and applications in risk-aware autonomous driving using latent features. In 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 4182–4189. IEEE, 2022

2022
[26]

Rescue robots at earthquake-hit mirandola, italy: A field report

Geert-Jan M Kruijff, Fiora Pirri, Mario Gianni, Panagi- otis Papadakis, Matia Pizzoli, Arnab Sinha, Viatcheslav Tretyakov, Thorsten Linder, Emanuele Pianese, Salvatore Corrao, et al. Rescue robots at earthquake-hit mirandola, italy: A field report. In 2012 IEEE international s...

2012
[27]

Deployment of ground and aerial robots in earthquake-struck amatrice in italy (brief report)

Ivana Kruijff-Korbayov ´a, Luigi Freda, Mario Gianni, Valsamis Ntouskos, V ´aclav Hlav ´aˇc, Vladim ´ır Kubelka, Erik Zimmermann, Hartmut Surmann, Kresimir Dulic, Wolfgang Rottner, et al. Deployment of ground and aerial robots in earthquake-struck amatrice in italy (brief repo...

2016
[28]

Learning terrain-aware kinodynamic model for autonomous off-road rally driving with model predictive path integral control

Hojin Lee, Taekyung Kim, Jungwi Mun, and Wonsuk Lee. Learning terrain-aware kinodynamic model for autonomous off-road rally driving with model predictive path integral control. IEEE Robotics and Automation Letters, 2023

2023
[29]

Strategic evacua- tion for hurricanes and regional events with and without autonomous vehicles

Jooyong Lee and Kara M Kockelman. Strategic evacua- tion for hurricanes and regional events with and without autonomous vehicles. Transportation research record , 2675(9):1398–1409, 2021

2021
[30]

Learning to walk from three minutes of real- world data with semi-structured dynamics models

Jacob Levy, Tyler Westenbroek, and David Fridovich- Keil. Learning to walk from three minutes of real- world data with semi-structured dynamics models. In 8th Annual Conference on Robot Learning , 2024

2024
[31]

Nonlinear observer-based adaptive control of ground vehicles with uncertainty estimation

Cuauht ´emoc Acosta L´ua, Domenico Bianchi, and Stefano Di Gennaro. Nonlinear observer-based adaptive control of ground vehicles with uncertainty estimation. Journal of the Franklin Institute , 360(18):14175–14189, 2023

2023
[32]

MAGIC-VFM-meta-learning adaptation for ground in- teraction control with visual foundation models

Elena Sorina Lupu, Fengze Xie, James A Preiss, Jedidiah Alindogan, Matthew Anderson, and Soon-Jo Chung. MAGIC-VFM-meta-learning adaptation for ground in- teraction control with visual foundation models. IEEE Transactions on Robotics , 2024

2024
[33]

Terrainnet: Visual modeling of complex terrain for high-speed, off-road navigation

Xiangyun Meng, Nathan Hatch, Alexander Lam- bert, Anqi Li, Nolan Wagener, Matthew Schmittle, JoonHo Lee, Wentao Yuan, Zoey Chen, Samuel Deng, et al. Terrainnet: Visual modeling of complex terrain for high-speed, off-road navigation. arXiv preprint arXiv:2303.15771, 2023

2023 arXiv
[34]

Ensemble gaussian processes for adaptive autonomous driving on multi-friction surfaces

Tom ´aˇs Nagy, Ahmad Amine, Truong X Nghiem, Ugo Rosolia, Zirui Zang, and Rahul Mangharam. Ensemble gaussian processes for adaptive autonomous driving on multi-friction surfaces. IFAC-PapersOnLine, 56(2):494– 500, 2023

2023
[35]

Autonomy for space robots: Past, present, and future

Issa AD Nesnas, Lorraine M Fesq, and Richard A V olpe. Autonomy for space robots: Past, present, and future. Current Robotics Reports , 2(3):251–263, 2021

2021
[36]

Graph-based multi-sensor fusion for consistent local- ization of autonomous construction robots

Julian Nubert, Shehryar Khattak, and Marco Hutter. Graph-based multi-sensor fusion for consistent local- ization of autonomous construction robots. In 2022 International Conference on Robotics and Automation (ICRA), pages 10048–10054. IEEE, 2022

2022
[37]

Neural-fly enables rapid learning for agile flight in strong winds

Michael O’Connell, Guanya Shi, Xichen Shi, Kamyar Azizzadenesheli, Anima Anandkumar, Yisong Yue, and Soon-Jo Chung. Neural-fly enables rapid learning for agile flight in strong winds. Science Robotics , 7(66): eabm6597, 2022

2022
[38]

Cahsor: Competence-aware high-speed off-road ground navigation in se (3)

Anuj Pokhrel, Aniket Datar, Mohammad Nazeri, and Xuesu Xiao. Cahsor: Competence-aware high-speed off-road ground navigation in se (3). arXiv preprint arXiv:2402.07065, 2024

2024 arXiv
[39]

Control-oriented meta-learning

Spencer M Richards, Navid Azizan, Jean-Jacques Slo- tine, and Marco Pavone. Control-oriented meta-learning. The International Journal of Robotics Research , 42(10): 777–797, 2023

2023
[40]

Continuous vehicle slip model identification on changing terrains

Forrest Rogers-Marcovitz, Neal Seegmiller, and Alonzo Kelly. Continuous vehicle slip model identification on changing terrains. In RSS 2012 Workshop on Long-term Operation of Autonomous Robotic Systems in Changing Environments, 2012

2012
[41]

Occam: Online continuous controller adaptation with meta-learned models

Hersh Sanghvi, Spencer Folk, and Camillo Jose Taylor. Occam: Online continuous controller adaptation with meta-learned models. In 8th Annual Conference on Robot Learning, 2024

2024
[42]

Metaverse: Meta-learning traversabil- ity cost map for off-road navigation

Junwon Seo, Taekyung Kim, Seongyong Ahn, and Kiho Kwak. Metaverse: Meta-learning traversabil- ity cost map for off-road navigation. arXiv preprint arXiv:2307.13991, 2023

2023 arXiv
[43]

Meta- adaptive nonlinear control: Theory and algorithms

Guanya Shi, Kamyar Azizzadenesheli, Michael O’Connell, Soon-Jo Chung, and Yisong Yue. Meta- adaptive nonlinear control: Theory and algorithms. Advances in Neural Information Processing Systems , 34:10013–10025, 2021

2021
[44]

Indirect adaptive control using neural network and discrete extended kalman filter for wheeled mobile robot

Mohammed Yousri Silaa, Aissa Bencherif, and Oscar Barambones. Indirect adaptive control using neural network and discrete extended kalman filter for wheeled mobile robot. In Actuators, volume 13, page 51. MDPI, 2024

2024
[45]

Rapidly adaptable legged robots via evolutionary meta- learning

Xingyou Song, Yuxiang Yang, Krzysztof Choromanski, Ken Caluwaerts, Wenbo Gao, Chelsea Finn, and Jie Tan. Rapidly adaptable legged robots via evolutionary meta- learning. in 2020 ieee. In RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 3769– 3776, 2020

2020
[46]

Autonomous robotics is driving perseverance rover’s progress on mars

Vandi Verma, Mark W Maimone, Daniel M Gaines, Raymond Francis, Tara A Estlin, Stephen R Kuhn, Gregg R Rabideau, Steve A Chien, Michael M McHenry, Evan J Graser, et al. Autonomous robotics is driving perseverance rover’s progress on mars. Science Robotics, 8(80):eadi3099, 2023

2023
[47]

Deep meta-learning energy-aware path planner for unmanned ground vehicles in unknown terrains

Marco Visca, Roger Powell, Yang Gao, and Saber Fal- lah. Deep meta-learning energy-aware path planner for unmanned ground vehicles in unknown terrains. IEEE access, 10:30055–30068, 2022

2022
[48]

Low frequency sampling in model predictive path integral control

Bogdan Vlahov, Jason Gibson, David D Fan, Patrick Spieler, Ali-akbar Agha-mohammadi, and Evangelos A Theodorou. Low frequency sampling in model predictive path integral control. IEEE Robotics and Automation Letters, 2024

2024
[49]

Pay attention to how you drive: Safe and adaptive model- based reinforcement learning for off-road driving

Sean J Wang, Honghao Zhu, and Aaron M Johnson. Pay attention to how you drive: Safe and adaptive model- based reinforcement learning for off-road driving. In 2024 IEEE International Conference on Robotics and Automation (ICRA), pages 16954–16960. IEEE, 2024

2024
[50]

Aggressive driving with model predictive path integral control

Grady Williams, Paul Drews, Brian Goldfain, James M Rehg, and Evangelos A Theodorou. Aggressive driving with model predictive path integral control. In 2016 IEEE International Conference on Robotics and Automation (ICRA), pages 1433–1440. IEEE, 2016

2016
[51]

Safe deep policy adaptation

Wenli Xiao, Tairan He, John Dolan, and Guanya Shi. Safe deep policy adaptation. In 2024 IEEE International Conference on Robotics and Automation (ICRA) , pages 17286–17292. IEEE, 2024

2024
[52]

Learning fast adaptation with meta strategy optimization

Wenhao Yu, Jie Tan, Yunfei Bai, Erwin Coumans, and Sehoon Ha. Learning fast adaptation with meta strategy optimization. IEEE Robotics and Automation Letters , 5 (2):2950–2957, 2020

2020

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.

[1] [1]

Online system identification and calibration of dynamic models for autonomous ground vehicles

Sina Aghli and Christoffer Heckman. Online system identification and calibration of dynamic models for autonomous ground vehicles. In 2018 IEEE International Conference on Robotics and Automation (ICRA) , pages 4933–4939. IEEE, 2018

work page 2018

[2] [2]

Learning and prediction of slip from visual information

Anelia Angelova, Larry Matthies, Daniel Helmick, and Pietro Perona. Learning and prediction of slip from visual information. Journal of Field Robotics , 24(3):205–231, 2007

work page 2007

[3] [3]

Few-shot semantic learning for robust multi-biome 3d seman- tic mapping in off-road environments

Deegan Atha, Xianmei Lei, Shehryar Khattak, Anna Sabel, Elle Miller, Aurelio Noca, Grace Lim, Jeffrey Edlund, Curtis Padgett, and Patrick Spieler. Few-shot semantic learning for robust multi-biome 3d seman- tic mapping in off-road environments. arXiv preprint arXiv:2411.06632, 2024

arXiv 2024

[4] [4]

Vehicle modeling and state estimation for autonomous driving in terrain

Tabish Badar, Juha Backman, and Arto Visala. Vehicle modeling and state estimation for autonomous driving in terrain. Control Engineering Practice, 152:106046, 2024

work page 2024

[5] [5]

Adaptive Meta-Learning for Identification of Rover-Terrain Dynamics

Somrita Banerjee, James Harrison, P Michael Furlong, and Marco Pavone. Adaptive meta-learning for iden- tification of rover-terrain dynamics. arXiv preprint arXiv:2009.10191, 2020

work page Pith review arXiv 2009

[6] [6]

Autonomous vehicles and intelligent automa- tion: Applications, challenges, and opportunities

Gourav Bathla, Kishor Bhadane, Rahul Kumar Singh, Rajneesh Kumar, Rajanikanth Aluvalu, Rajalakshmi Kr- ishnamurthi, Adarsh Kumar, RN Thakur, and Shakila Basheer. Autonomous vehicles and intelligent automa- tion: Applications, challenges, and opportunities. Mobile Information Systems, 2022(1):7632892, 2022

work page 2022

[7] [7]

Model-based meta-reinforcement learning for flight with suspended payloads

Suneel Belkhale, Rachel Li, Gregory Kahn, Rowan McAllister, Roberto Calandra, and Sergey Levine. Model-based meta-reinforcement learning for flight with suspended payloads. IEEE Robotics and Automation Letters, 6(2):1471–1478, 2021

work page 2021

[8] [8]

Risk-aware off-road navigation via a learned speed distribution map

Xiaoyi Cai, Michael Everett, Jonathan Fink, and Jonathan P How. Risk-aware off-road navigation via a learned speed distribution map. In 2022 IEEE/RSJ Inter- national Conference on Intelligent Robots and Systems (IROS), pages 2931–2937. IEEE, 2022

work page 2022