Dynamics Modeling using Visual Terrain Features for High-Speed Autonomous Off-Road Driving

Anoushka Alavilli; Erica Tevere; Evangelos A. Theodorou; Jason Gibson; Patrick Spieler

REVIEW 2 major objections 6 minor 23 references

Dynamics Modeling using Visual Terrain Features for High-Speed Autonomous Off-Road Driving

T0 review · 2 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read Feeding compressed visual terrain features from DINOv2 into a hybrid dynamics model cuts mean 5-second trajectory prediction loss by about 10%, with most of the gain an 8% reduction in longitudinal velocity error at the horizon.

desk verdict A solid empirical contribution to vision-informed off-road dynamics, but the same-log train/test split leaves the headline 10% gain short of proof of generalization. read the letter →

arxiv 2412.00581 v1 pith:EWZAA3LI submitted 2024-11-30 cs.RO

Jason Gibson , Anoushka Alavilli , Erica Tevere , Evangelos A. Theodorou , Patrick Spieler This is my paper · ORCID

classification cs.RO

keywords visualfoundationmodelsoff-roaddrivingvehicledynamicsmodelpredictivecontrolterrainfeaturecompressionDINOv2hybridtrajectoryprediction

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 terrain appearance seen by a camera can predict how a vehicle's dynamics will change during high-speed off-road driving, and that a small learned encoder can compress the visual features enough for real-time planning. The authors build a hybrid dynamics model with physics-based bicycle equations plus an LSTM correction network, and give it, at each wheel, a low-dimensional encoding of DINOv2 visual features of the ground. On roughly two million 5-second trajectories from aggressive driving in four environments, the vision-informed model reduces mean summed prediction loss by about 10 percent relative to the same model without vision, with the largest single gain an 8 percent reduction in the mean longitudinal velocity error at the end of the horizon. A reader should care because if this holds, planners can look ahead at terrain and anticipate traction losses before they happen, instead of reacting after the vehicle slips.

What carries the argument

The load-bearing mechanism is a distance-independent feature encoder $\zeta_E$: a fully connected network with hidden sizes [64, 32] and tanh activations that maps a 40-dimensional PCA projection of DINOv2 ground-pixel features to a smaller encoding (varied between 1 and 8 dimensions) for each wheel. It is trained end-to-end with an LSTM force-compensation network $\zeta_\mu$, whose output adjusts the parametric bicycle-model force as $\dot{\hat{x}}_t = h(F_t + \zeta_\mu(\hat{x}_t, u_t, y_t, F_t, \zeta_E(\hat{y}_f)), \hat{x}_t)$. During training, features are randomly drawn from seven 10 m distance buckets plus a hindsight map, together with a missing-data flag, so the encoder learns a feature basis that stays valid when the runtime 2D map contains features projected from different distances and partially occluded. The PCA basis itself comes from 175 manually selected images masked to ground pixels.

What would settle it

Train the same vision-informed hybrid model with the visual feature input replaced by a fixed constant or by random noise drawn at the same locations, keeping every other part of the training procedure identical; if the roughly 10 percent summed-loss reduction and 8 percent longitudinal-velocity reduction do not disappear or shrink, the claimed effect is not caused by the terrain features themselves.

Watch

Extended reading notes

Core claim

The central discovery is that visual terrain features carry predictively useful information about terradynamics that the same learned dynamics model cannot otherwise extract: adding compressed DINOv2 ground-patch features as inputs to each wheel's force compensation lowers 5-second trajectory prediction error by roughly 10%, with about 8% of that coming from improved longitudinal velocity prediction at the horizon. The paper further shows that a naive feature-injection model trained only on hindsight features fails once features are evaluated at the projection distances used at runtime, whereas a feature encoder trained end-to-end with randomized distance buckets maintains the improvement from 0 to 30 m, where planning predictions occur.

Load-bearing premise

The whole result rests on the assumption that the compressed, ground-masked DINOv2 features still carry enough information about how the terrain will change grip, slip, and rolling resistance that a learned model can exploit it, despite the features varying with viewing distance and being missing for large portions of the terrain.

Editorial extensions

If this is right

Planners using this dynamics model can anticipate terrain-induced changes in traction, slip, and rolling resistance from vision alone, reducing the need for online model adaptation.
The roughly 8% lower longitudinal velocity error at 5 s also improves yaw prediction, because the parametric yaw rate is heavily dependent on speed.
Because the encoder output is tiny (1 to 8 dimensions per wheel), mapping visual features into the 2D terrain map becomes computationally tractable at runtime.
Performance is robust to the encoder output size and to the number of PCA components, so memory and computation can be traded away without losing the benefit.

Reading between the lines

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

A strong next test would be evaluating the trained model on a held-out terrain environment not present in the training logs, since the paper's train/test split is by trajectory and not by environment.
The useful range stops around 30 m where LiDAR occlusion invalidates many features, so fusing temporal or hindsight map updates into the runtime map could push the benefit farther out.
One could probe what the distance-independent encoder learns by clustering its encodings and checking whether clusters correspond to physical terrain classes such as sand, grass, or ruts.

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

Jason Gibson et al. report that compressing DINOv2 terrain features with an encoder trained end-to-end with a hybrid dynamics model improves 5-second trajectory prediction by about 10% over the same model without visual features, on a large dataset of aggressive off-road driving. The main new piece is the distance-bucket randomization: they train on features projected from different distances (and with a hindsight map), which makes the compressed representation robust to projection distance and occlusion. That is a real practical problem for runtime mapping, and the ablations in Fig. 5b show the approach beats both naive feature input and the featureless baseline at 0-30 m. The paper also does useful sensitivity analysis on compression dimension and PCA size, and it is honest about failing beyond 30 m and about the large occlusion percentages.

The soft spots are real but not fatal. The biggest one is the evaluation protocol: the test set is drawn from the same logs as training, just non-overlapping trajectories. That means the ~10% gain could partly reflect the learned encoder memorizing appearance of those four environments rather than learning terrain dynamics that generalize. The stress-test note has it right. A held-out environment or at least a temporally disjoint split would be needed to know. Relatedly, there are no confidence intervals or significance tests; the boxplot whiskers are the only spread shown, and the mean differences are small (about 5.03 vs 5.47 at 0-10 m, about 8% vx reduction). The PCA basis comes from 175 manually selected images, which is a minor concern since they show insensitivity to npca. And everything is offline; there is no closed-loop planning result, so the claimed path to MPC is plausible but unproven.

I don't think the central hypothesis is broken. The comparison with and without visual features is honest, and the distance-robustness story is well motivated. But the paper oversells the general conclusion relative to what the same-log evaluation can support. It deserves a serious referee, mostly to push for a stronger evaluation split and error bars. I would not cite it in my own work until the generalization question is addressed, but it is worth putting in front of a reading group that cares about terrain-aware dynamics.

Referee Report

2 major / 6 minor

Summary. The paper presents a hybrid vehicle dynamics model for high-speed off-road driving. The model combines a parametric bicycle model with an LSTM residual; the residual is conditioned on compressed DINOv2 visual terrain features extracted at each wheel location. A fully-connected feature encoder is trained end-to-end with the dynamics model, and a projection-distance randomization scheme is introduced to make the encoder robust to features gathered at different ranges. The method is evaluated offline on roughly 2 million 5-second trajectories collected from four environments, reporting about a 10% reduction in mean summed loss relative to a no-vision baseline, with most of the gain in longitudinal velocity prediction. Sensitivity analyses cover compression dimension and PCA basis size, and the distance-randomized model is shown to outperform the baseline when features come from forward-projected maps up to 30 m ahead.

Significance. Strengths: the dataset is large and from real aggressive off-road driving; the paper ships explicit comparisons between direct and compressed feature inputs and between hindsight and forward-projected features; the distance-bucket occlusion statistics are a useful characterization of the perception problem; and the proposed runtime architecture (compress before mapping) is practical. If the prediction gains survive a stricter evaluation, this would be a solid step toward using foundation-model visual features inside MPC rollouts. The main weakness is empirical: the test split is from the same logs as training, with no held-out environment or temporal gap, and there are no confidence intervals, so the headline 10% is not yet established as a general gain.

major comments (2)

[Section VI (evaluation protocol)] The statement that 'training and test sets are derived from the same logs but have no overlap' is the load-bearing evaluation point, and as it stands it is too weak to support the generalization claim. Because all four environments are represented in both sets and no temporally or spatially disjoint split is reported, the reported ~10% reduction in mean summed loss and ~8% reduction in vx error could reflect memorization of environment-specific visual appearance rather than learning a general terrain-to-dynamics mapping. The end-to-end-trained feature encoder makes this channel especially plausible, since DINOv2 features are known to encode lighting and scene identity. Please add a leave-one-environment-out evaluation (or at least a split by recording log/date) with per-environment error for the DC model and the baseline, and report the variance across the split. Without this, the central claim of generalization to new terrain is not supported.
[Section VI-B, Fig. 5b] The runtime-relevant improvement is distance-dependent and small, and the paper provides no uncertainty quantification. In Fig. 5b, the DC mean position error is about 5.03 at 0m/10m versus 5.47 for the baseline B, 5.11 at 20m, 5.30 at 30m, and 5.55 at 40m (no improvement). The differences at 20-30m are small relative to the displayed spread of the boxplots, and no confidence intervals, significance tests, or repeated-seed statistics are given anywhere in Section VI. Please provide bootstrap confidence intervals over trajectories (or multiple training runs) for the comparison DC versus B at each distance bucket, and state the effect size. This is necessary to judge whether the proposed approach truly helps at deployment-relevant distances beyond 10 m.

minor comments (6)

[Section VI-A] There is a typo 'compared as compared to'; it should read 'compared to'.
[Section V-A] The text says '7 different distance buckets' but the occlusion list in the same paragraph contains 8 entries (hindsight plus -20, -10, ..., 40 m). Please reconcile the count or define the buckets explicitly.
[Section VI] 'Initialization LSTMs have 20 hidden layers' should say 'hidden units' (or 'hidden size'); 20 layers is almost certainly not intended.
[Figures 4 and 5] The y-axis label 'Distance (m)' is ambiguous; please state explicitly that it is the Euclidean position error at the 5 s horizon, and define the 'mean summed loss' metric used for the 10% claim (e.g., sum of squared errors over which state components and time steps).
[Section IV] In Eq. (5), the symbolic input y_t is not defined in the text; please state that it contains the elevation normals and visual features queried at the wheel locations.
[Section IV] The sentence 'For the first three modules, we model the delay in actuation or RPM as a state' is vague; explicitly define the states x_rpm, x_br, and delta, or give a reference to the equation where they are introduced.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity; the central visual-feature improvement is an empirical ablation, not a derivation that assumes its own conclusion.

full rationale

The paper's central claim—that DINOv2-derived terrain features improve 5-second dynamics prediction—is tested by comparing hybrid dynamics models with and without those features and evaluating on a test set of non-overlapping trajectories from the same logs (Section VI). The end-to-end training of the feature encoder zeta_E together with the dynamics loss (Eq. 6) is standard supervised feature learning: since the evaluation is on disjoint held-out trajectories, improved test error is not guaranteed by construction. There is no equation or definition in the paper in which the predicted quantity equals a fitted quantity, nor is any parameter fitted to a subset and then reported as a prediction of that same subset. The self-citations [7] and [20] are used to specify the dynamics-learning setup and the MPPI variant, but they are not invoked as a uniqueness theorem or as the evidence for the visual-feature benefit; that evidence is the ablation in Figures 4 and 5. The same-log/no-held-out-environment protocol is a legitimate generalization-risk concern, but is not circularity. No self-definitional, fitted-input-called-prediction, imported-uniqueness, or ansatz-by-citation pattern is present.

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

The paper does not introduce new physical entities or forces. The main free parameters are the compression dimensions and the PCA basis, which are design choices rather than fitted physical constants. The key domain assumptions are that DINOv2 features are dynamically informative after compression and that the bicycle model is an adequate base. The lack of a cross-environment evaluation split weakens the generalization claim.

free parameters (4)

PCA basis size (npca) = 40
Chosen as a trade-off between memory and model quality. The paper shows performance is not highly sensitive to this value, but it is a hand-selected hyperparameter.
Final compression dimension (nencoder) = 4 (in C4)
The output dimension of the feature encoder is varied in Fig. 4b and the paper uses nencoder=4 for the main model (C4). This is a design choice.
Number of distance buckets = 7
Features are collected from 7 distance buckets plus hindsight, spaced every 10m up to 40m. This is a hand-chosen design parameter.
PCA basis training images = 175
A set of 175 images were manually selected to compute the PCA basis. This is a manual selection that could bias the feature space.

assumptions (4)

domain assumption DINOv2 visual features, masked to ground pixels, contain information about terrain properties that affect vehicle dynamics.
The central premise of the paper. Invoked in Section V, where the feature encoder is trained to compress the visual features into a dynamics-relevant subspace.
domain assumption The bicycle dynamics model with Pacejka tire curves is an adequate parametric base for the vehicle dynamics.
The parametric model in Eq. (1)-(4) is drawn from prior work [10] and assumed to capture the nominal dynamics well enough that a neural compensation term is sufficient.
domain assumption The 0.2m voxel map and nearest-neighbor filling of missing data preserve the terrain information relevant to dynamics.
The mapping pipeline in Section III-B retains the lowest valid data point in each vertical voxel column and fills gaps with the nearest neighbor within 0.4m, which may lose or distort terrain features.
domain assumption Training and test splits from the same logs generalize to new environments.
The paper evaluates on a held-out test set from the same logs; no cross-environment split is described. The claim of generalization across environments is therefore not directly supported.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Dynamics Modeling using Visual Terrain Features for High-Speed Autonomous Off-Road Driving." pith.science (2026). https://pith.science/paper/EWZAA3LI

@misc{pith2026241200581,
  author       = {Pith},
  title        = {Pith review of: Dynamics Modeling using Visual Terrain Features for High-Speed Autonomous Off-Road Driving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EWZAA3LI}},
  note         = {Machine review of arXiv:2412.00581}
}

read the original abstract

Rapid autonomous traversal of unstructured terrain is essential for scenarios such as disaster response, search and rescue, or planetary exploration. As a vehicle navigates at the limit of its capabilities over extreme terrain, its dynamics can change suddenly and dramatically. For example, high-speed and varying terrain can affect parameters such as traction, tire slip, and rolling resistance. To achieve effective planning in such environments, it is crucial to have a dynamics model that can accurately anticipate these conditions. In this work, we present a hybrid model that predicts the changing dynamics induced by the terrain as a function of visual inputs. We leverage a pre-trained visual foundation model (VFM) DINOv2, which provides rich features that encode fine-grained semantic information. To use this dynamics model for planning, we propose an end-to-end training architecture for a projection distance independent feature encoder that compresses the information from the VFM, enabling the creation of a lightweight map of the environment at runtime. We validate our architecture on an extensive dataset (hundreds of kilometers of aggressive off-road driving) collected across multiple locations as part of the DARPA Robotic Autonomy in Complex Environments with Resiliency (RACER) program. https://www.youtube.com/watch?v=dycTXxEosMk

Figures

Figures reproduced from arXiv: 2412.00581 by the authors.

**Figure 2.** Terrain geometries and properties vary significantly across the [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Left: A forward-facing image of size R960×594×3 (in RGB). Right: VFM output of size R68×42×384, where each 14×14 pixel patch results in one feature vector of size R1×384. DINOv2 features from ground regions undergo PCA, and the first three components are visualized in RGB. The result effectively segments on- and off-trail terrain. 1) Visual Features: As the vehicle drives, images are captured from four RGB cameras f… view at source ↗

Figures from the paper (2 more)

**Figure 4.** Figure 4: Distance error of models at 5s using best features in hindsight, B is a no feature baseline, DF is directly inputting features into the network, and C is compressing features. The model CF in 4a, C 4 in 4b, C 40 in 4c are all the same and axis are kept consistent betwe…

**Figure 5.** Figure 5: Distance error of at 5s models on features at varying projection distances. The model DF and CF are kept consistent from Fig. 4a. DC is our proposed distance independent approach. Whiskers are defined as ±1.5IQR and given by the values with arrows, the green line defin…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

23 extracted references · 17 canonical work pages

[1]

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
[2]

EVORA: deep evidential traversability learning for risk-aware off-road autonomy, 2024

Xiaoyi Cai et al. EVORA: deep evidential traversability learning for risk-aware off-road autonomy, 2024

work page 2024
[3]

Identifying terrain physical parameters from vision – towards physical-parameter-aware locomotion and navigation, 2024

Jiaqi Chen et al. Identifying terrain physical parameters from vision – towards physical-parameter-aware locomotion and navigation, 2024

work page 2024
[4]

Mem: Multi-modal elevation mapping for robotics and learn- ing

Gian Erni, Jonas Frey, Takahiro Miki, Matias Mattamala, and Marco Hutter. Mem: Multi-modal elevation mapping for robotics and learn- ing. In 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 11011–11018, 2023

work page 2023
[5]

Roadrunner – learning traversability estimation for autonomous off-road driving

Jonas Frey et al. Roadrunner – learning traversability estimation for autonomous off-road driving. arxiv.org:2402.19341, 2024

arXiv 2024
[6]

Gasparino et al

Mateus V . Gasparino et al. WayFAST: navigation with predictive traversability in the field. IEEE Robotics and Automation Letters , 7(4):10651–10658, October 2022

work page 2022
[7]

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

Jason Gibson et al. A multi-step dynamics modeling framework for autonomous driving in multiple environments. arxiv.org:2305.02241, 2023

arXiv 2023
[8]

Terrain-aware path planning and map update for Mars sample return mission

Gabrielle Hedrick, Nicholas Ohi, and Yu Gu. Terrain-aware path planning and map update for Mars sample return mission. IEEE Robotics and Automation Letters , 5(4):5181–5188, 2020

work page 2020

Show all 23 references

[9]

V-STRONG: visual self-supervised traversability learning for off-road navigation, 2024

Sanghun Jung, JoonHo Lee, Xiangyun Meng, Byron Boots, and Alexander Lambert. V-STRONG: visual self-supervised traversability learning for off-road navigation, 2024

2024
[10]

Zeilinger

Juraj Kabzan, Lukas Hewing, Alexander Liniger, and Melanie N. Zeilinger. Learning-based model predictive control for autonomous racing. IEEE Robotics and Automation Letters , 4(4):3363–3370, 2019

2019
[11]

Kingma and Jimmy Ba

Diederik P. Kingma and Jimmy Ba. Adam: A method for stochastic optimization, 2017

2017
[12]

Gaussian process based model predictive control for overtaking scenarios at highway curves

Wenjun Liu, Yulin Zhai, Guang Chen, and Alois Knoll. Gaussian process based model predictive control for overtaking scenarios at highway curves. In 2022 IEEE Intelligent V ehicles Symposium (IV) , pages 1161–1167, 2022

2022
[13]

Preiss, Jedidiah Alin- dogan, Matthew Anderson, and Soon-Jo Chung

Elena Sorina Lupu, Fengze Xie, James A. Preiss, Jedidiah Alin- dogan, Matthew Anderson, and Soon-Jo Chung. MAGIC-VFM: Meta-learning adaptation for ground interaction control with Visual Foundation Models, 2024

2024
[14]

Two years of visual odometry on the Mars exploration rovers

Mark Maimone, Yang Cheng, and Larry Matthies. Two years of visual odometry on the Mars exploration rovers. Journal of Field Robotics , 24(3):169–186, 2007

2007
[15]

Wild visual navigation: Fast traversabil- ity learning via pre-trained models and online self-supervision

Mat ´ıas Mattamala et al. Wild visual navigation: Fast traversabil- ity learning via pre-trained models and online self-supervision. arxiv.org:2404.07110, 2024

2024 arXiv
[16]

DINOv2: learning robust visual features without supervision, 2024

Maxime Oquab et al. DINOv2: learning robust visual features without supervision, 2024

2024
[17]

Assessing Mars curiosity rover wheel damage

Arturo Rankin, Nikunj Patel, Evan Graser, Jiun-Kai Freddy Wang, and Kimberly Rink. Assessing Mars curiosity rover wheel damage. In IEEE Aerospace Conference (AERO) , pages 1–19, 2022

2022
[18]

Predicting semantic map representations from images using pyramid occupancy networks

Thomas Roddick and Roberto Cipolla. Predicting semantic map representations from images using pyramid occupancy networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 11138–11147, 2020

2020
[19]

Semantic terrain classification for off-road autonomous driving

Amirreza Shaban, Xiangyun Meng, JoonHo Lee, Byron Boots, and Dieter Fox. Semantic terrain classification for off-road autonomous driving. In Aleksandra Faust, David Hsu, and Gerhard Neumann, editors, Proceedings of the 5th Conference on Robot Learning , volume 164 of Proceedin...

2022
[20]

Fan, Patrick Spieler, Ali- akbar Agha-mohammadi, and Evangelos A

Bogdan Vlahov, Jason Gibson, David D. Fan, Patrick Spieler, Ali- akbar Agha-mohammadi, and Evangelos A. Theodorou. Low fre- quency sampling in model predictive path integral control. IEEE Robotics and Automation Letters , 9(5):4543–4550, May 2024

2024
[21]

Theodorou

Bogdan Vlahov, Jason Gibson, Manan Gandhi, and Evangelos A. Theodorou. MPPI-Generic: a CUDA library for stochastic optimiza- tion, 2024

2024
[22]

Model predictive path integral control using covariance variable importance sampling, 2015

Grady Williams, Andrew Aldrich, and Evangelos Theodorou. Model predictive path integral control using covariance variable importance sampling, 2015

2015
[23]

Contrastive label disambiguation for self- supervised terrain traversability learning in off-road environments

Hanzhang Xue et al. Contrastive label disambiguation for self- supervised terrain traversability learning in off-road environments. arxiv.org:2307.02871, 2023

2023 arXiv

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.

[1] [1]

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

[2] [2]

EVORA: deep evidential traversability learning for risk-aware off-road autonomy, 2024

Xiaoyi Cai et al. EVORA: deep evidential traversability learning for risk-aware off-road autonomy, 2024

work page 2024

[3] [3]

Identifying terrain physical parameters from vision – towards physical-parameter-aware locomotion and navigation, 2024

Jiaqi Chen et al. Identifying terrain physical parameters from vision – towards physical-parameter-aware locomotion and navigation, 2024

work page 2024

[4] [4]

Mem: Multi-modal elevation mapping for robotics and learn- ing

Gian Erni, Jonas Frey, Takahiro Miki, Matias Mattamala, and Marco Hutter. Mem: Multi-modal elevation mapping for robotics and learn- ing. In 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 11011–11018, 2023

work page 2023

[5] [5]

Roadrunner – learning traversability estimation for autonomous off-road driving

Jonas Frey et al. Roadrunner – learning traversability estimation for autonomous off-road driving. arxiv.org:2402.19341, 2024

arXiv 2024

[6] [6]

Gasparino et al

Mateus V . Gasparino et al. WayFAST: navigation with predictive traversability in the field. IEEE Robotics and Automation Letters , 7(4):10651–10658, October 2022

work page 2022

[7] [7]

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

Jason Gibson et al. A multi-step dynamics modeling framework for autonomous driving in multiple environments. arxiv.org:2305.02241, 2023

arXiv 2023

[8] [8]

Terrain-aware path planning and map update for Mars sample return mission

Gabrielle Hedrick, Nicholas Ohi, and Yu Gu. Terrain-aware path planning and map update for Mars sample return mission. IEEE Robotics and Automation Letters , 5(4):5181–5188, 2020

work page 2020