Explore Where It Matters: Targeted Exploration for Efficient Rover Adaptation

Jacob Levy; Arthur Bouton; Tristan D. Hassler; Hari Nayar; David Fridovich-Keil

cs.ROOct 28, 2026

Explore Where It Matters: Targeted Exploration for Efficient Rover Adaptation

Authors: Jacob Levy, Arthur Bouton, Tristan D. Hassler, Hari Nayar, David Fridovich-Keil

Organizations: Department of Aerospace Engineering, University of Texas at Austin · Jet Propulsion Laboratory, California Institute of Technology

Abstract

Future planetary rovers will traverse steep, deformable terrain where slip, sinkage, and terrain deformation can threaten mobility. Although simulation can capture some wheel-soil effects, unmodeled deformable-terrain phenomena create sim-to-real gaps requiring deployment-time adaptation. Yet, rover operations leave little room for extensive trial-and-error. We introduce TERRA (Targeted Exploration for Reward-Relevant Adaptation), an online adaptation framework that efficiently seeks deployment data exposing task-relevant model errors and corrects them. TERRA adapts a simulation-trained world model with a reward-weighted prediction loss and trains a loss predictor to estimate where future performance-relevant errors are likely to occur, using this signal to guide exploration during MPC. In simulation, TERRA reaches approximately 90% slope-climbing success within five trials under severe out-of-distribution pretraining conditions, outperforming exploration baselines in reward, completion time, and energy to first success. On the ERNEST rover, TERRA recovers coordinated active-suspension locomotion on a deformable Mars-simulant slope within three to five real-world trials. Project website: https://terra-explore.github.io

Figures & tables

Figure 1: TERRA rapidly adapts to an out-of-distribution deformable slope. No-Slope pretraining never includes slopes, while Hard-Ground never requires coordinated use of the Active Gimbal Suspension (AGS). Both initially entrap the ERNEST rover, but TERRA learns to climb using its AGS within three to five trials.

Figure 2: AGS actuation in sim.

Figure 3: TERRA overview. A simulation-pretrained world model rolls out candidate action sequences for sampling-based planning, while a loss predictor estimates where future performance-relevant dynamics errors are likely to occur. The planner combines predicted task reward with this targeted exploration signal, executes the selected action on the rover, and updates both models with the resulting deployment data.

Figure 4: Simulated slope-climbing adaptation under two severe out-of-distribution pretraining conditions. Across 100 random seeds, TERRA reaches approximately 90% success within five deployment trials in both No-Slope and Hard-Ground, while achieving higher task reward, shorter completion times, and lower energy to first success than exploration baselines. Curves show means with 95% bootstrapped confidence intervals.

	No-Slope		Hard-Ground
	SR $\uparrow$	Energy (kJ) $\downarrow$	SR $\uparrow$	Energy (kJ) $\downarrow$
TERRA	0.91	50.3 [47, 53]	0.89	51.6 [49, 54]
No expl.	0.85	56.5 [52, 61]	0.85	61.0 [57, 65]
No RWL	0.73	80.9 [74, 88]	0.86	60.4 [56, 65]
No expl. / RWL	0.61	78.7 [72, 86]	0.87	57.3 [53, 62]

Table 1: Ablations without exploration (expl.) and/or reward-weighted loss (RWL). Across 100 seeds, we report success rate (SR) at trial 5 and energy to first success with 95% bootstrapped CIs.

Figure 5: Adaptation metrics for Hard-Ground. Left: each method’s exploration signal evolution over deployment. Middle: lower prediction error alone does not necessarily yield better task performance. Trial-5 error and reward-gradient magnitude for TERRA and RND, showing that TERRA reduces error where the reward is most sensitive, while RND achieves lower aggregate error largely on reward-insensitive observation groups.

Figure 6: Real-world AGS yaw during Hard-Ground. TERRA adapts from little AGS use in Trial 1 to large alternating yaw motions by the successful Trial 3 climb.

Appendix figures & tables13 assets

Supplementary material from the paper’s appendix.

Appendix

Observation	Dimension
Pose, SE(2)	3
AGS Yaw and Roll	2
Body Roll and Pitch	2
Wheel Forces	12
Wheel Torques	12
Height Map Sampling	210

Appendix

Table 2: Observation space.

Action	Dimension
Commanded Forward Velocity	1
AGS Yaw Position Command	1
AGS Roll Position Command	1

Appendix

Table 3: Action space.

Term Name	Reward Term	Weight
Forward advance reward	$\displaystyle\phantom{-}\dfrac{x_{t+1}-x_{t}}{\Delta t}$	$50.0$
Lateral deviation penalty	$\displaystyle-y_{t+1}^{2}$	$2.5$
Bearing deviation penalty	$\displaystyle-\beta_{t+1}^{2}$	$1.0$
AGS Yaw penalty	$\displaystyle-q_{\mathrm{yaw},t+1}^{2}$	$1.25$
AGS Roll penalty	$\displaystyle-q_{\mathrm{roll},t+1}^{2}$	$10.0$
Yaw action penalty	$\displaystyle-a_{\mathrm{yaw},t}^{2}$	$1.0$

Appendix

Table 4: Task reward function,

R^{\mathtt{task}}

. Here,

x

and

y

are the forward and lateral displacements of the robot;

\beta

is the deviation from initial robot bearing; and

q_{\mathrm{yaw}}

and

q_{\mathrm{roll}}

are the AGS joint angles.

Parameter	Value
Discount factor	0.99
Soft target update factor	$5\text{\times}{10}^{-3}$
Policy update delay	2
Target policy regularization noise	$\mathcal{N}(0,\,0.05^{2})$
Bounds of the target policy regularization	0.2

Appendix

Table 5: Hyperparameters for RL with TD3.

Pretraining Condition	Ripples	Step	Flat	Slope
No-Slope	1024	1024	–	–
Hard-Ground	512	–	512	512

Appendix

Table 6: Number of parallel environments for each pretraining condition.

Initial rover pose
Parameter	Value / Range
Forward position ( $\mathrm{m}$ )	$[-1.0,\,0.0]$
Lateral position ( $\mathrm{m}$ )	$[-0.2,\,0.2]$
Bearing angle ( $\mathrm{\SIUnitSymbolDegree}$ )	$[-5.0,\,5.0]$
Soil Parameters
Soil density ( $\mathrm{kg}\text{\,}{\mathrm{m}}^{-3}$ )	$1600$

Appendix

Table 7: Common parameters during RL for both the No-Slope and Hard-Ground pretraining conditions. Individual values indicate fixed parameters and ranges indicate uniform random sampling within the provided ranges.

Parameter	Value / Range
Soil Parameters
Internal friction angle ( $\mathrm{\SIUnitSymbolDegree}$ )	$[30,\,40]$
Step
Episode length ( $\mathrm{s}$ )	$45$
Height ( $\mathrm{m}$ )	$[0.2,\,0.4]$
Roll ( $\mathrm{\SIUnitSymbolDegree}$ )	$[-1,\,1]$

Appendix

Table 8: Parameters during RL for No-Slope only. Individual values indicate fixed parameters and ranges indicate uniform random sampling within the provided ranges.

Soil Parameters
Parameter	Value / Range
Internal friction angle ( $\mathrm{\SIUnitSymbolDegree}$ )	$60$
Flat
Episode length ( $\mathrm{s}$ )	$30$
Roll ( $\mathrm{\SIUnitSymbolDegree}$ )	$[-5,\,5]$
Pitch ( $\mathrm{\SIUnitSymbolDegree}$ )	$[-5,\,5]$

Appendix

Table 9: Parameters during RL for Hard-Ground only. Individual values indicate fixed parameters and ranges indicate uniform random sampling within the provided ranges.

Parameter	Value
Embedding dimension	$128$
Encoder (observations) MLP hidden dims	$256,256$
All transformers MLP hidden size	$512$
Dynamics transformer layers	$4$
Dynamics transformer heads	$8$
Base policy transformer layers	$4$

Appendix

Table 10: World model and loss predictor architectural parameters.

Figure 7: Detailed world model and loss predictor architecture.

Parameter	Value
Candidate actions batch size	$1024$
Noised base policy actions batch size	$2$
Solver iterations	$10$
Initial action standard deviation	$[0.2,0.7,0.78]$
Minimum action standard deviation	$0.001$
Base policy action standard deviation	$[0.025,0.05,0.05]$

Appendix

Table 11: MPPI parameters.

Parameter	Value
LoRA rank	$64$
Ordinary prediction error weight, $c_{1}$	$0.04$
Reward-weighted prediction error weight, $c_{2}$	$0.005$
Loss bin thresholds, $\{\tau_{j}\}_{j=1}^{N}$	$\{0,0.05,0.1,0.2,0.4,0.8\}$
Class weights, $w$	$[0.5,0.75,1,5,10,20]$
Exploration reward strength, $\lambda$	$0.8$

Appendix

Table 12: TERRA parameters.

(E.1) Adding Reward-Weighted Adaptation to Baselines
	SR $\uparrow$	Reward $\uparrow$	Time ( $\mathrm{s}$ ) $\downarrow$	Energy ( $\mathrm{kJ}$ ) $\downarrow$	REL $\uparrow$
TERRA	0.91	2.46 [2.3, 2.6]	83.4 [82, 84]	50.3 [47, 53]	1.00
RND	0.88	2.46 [2.3, 2.6]	84.5 [83, 86]	53.1 [50, 57]	0.98
P2E	0.88	2.36 [2.2, 2.5]	84.8 [83, 86]	52.4 [50, 55]	0.97
RUNE	0.81	1.98 [1.8, 2.2]	87.3 [86, 89]	52.8 [50, 56]	0.90
Disagreement	0.80	2.32 [2.1, 2.5]	90.5 [88, 93]	59.2 [56, 63]	0.90

Appendix

Table 13: Additional experiments and ablations. Unless noted otherwise, No-Slope, Trial-5, 100 seeds; brackets show 95% bootstrap CIs. REL is the geometric mean performance ratio to TERRA across success rate, reward, completion time, and energy to first success;

\text{REL}<1

is worse than TERRA and

\text{REL}>1

is better.

Planetary Rover Robotics

See all 20 papers

Jun 5, 2026

Learning All-Terrain Locomotion for a Planetary Rover with Actively Articulated Suspension

This paper presents ERNEST, a four-wheeled planetary rover concept equipped with a two-degree-of-freedom Active Gimbal Suspension that combines yaw and roll actuation to enable wheel reconfiguration, steering, and active load redistribution. A single neural network controller, trained to track a desired path across challenging terrain, fully unlocks the capabilities of this actuated suspension system for autonomous obstacle negotiation. A reinforcement learning framework is developed using the high-fidelity DARTS simulation engine, which combines rigid-contact dynamics and Bekker-Wong terramechanics, enabling the emergence of locomotion strategies adapted to loose-soil conditions. To obtain a single unified controller across heterogeneous terrains, a policy consolidation strategy merges the experience of terrain-specialized agents into one neural network, eliminating the need for explicit terrain classification and controller switching. The resulting controller operates on a combination of proprioceptive and exteroceptive feedback, including sparse stereo-derived terrain elevation, chassis attitude, joint states, and force-torque measurements. Zero-shot transfer to the physical rover is achieved through domain randomization, sensor noise injection, and model-to-real system identification. Experimental results demonstrate autonomous traversal of rock fields, a Bickler trap (bump obstacle), a wheel-high step, sand ripples, and sandy slopes. On a 20° sandy slope, the learned controller reduces the cost of transport by 37% on dry sand despite the additional actuation, and achieves superior performance on wet sand where the passive suspension becomes completely immobilized. A video accompanying this paper is available at https://youtu.be/d684P5a3xMc

Arthur Bouton, Tristan D. Hasseler, Michael Paton +5

Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA 91109 USA · Center for Autonomy, University of Texas at Austin, Austin, TX 78712 USA · Space Systems Laboratory, University of Maryland, MD 20742 USA

Planetary Rover Robotics Off-Road Robot Navigation Robotic RL Robotics Terrain-Aware Robot Locomotion Vehicle Dynamics+3

Ask PDF

Mar 9, 2026

Proprioceptive Safe Active Navigation and Exploration for Planetary Environments

Deformable granular terrains introduce significant locomotion and immobilization risks in planetary exploration and are difficult to detect via remote sensing (e.g., vision). Legged robots can sense terrain properties through leg-terrain interactions during locomotion, offering a direct means to assess traversability in deformable environments. How to systematically exploit this interaction-derived information for navigation planning, however, remains underexplored. We address this gap by presenting PSANE, a Proprioceptive Safe Active Navigation and Exploration framework that leverages leg-terrain interaction measurements for safe navigation and exploration in unknown deformable environments. PSANE learns a traversability model via Gaussian Process regression to estimate and certify safe regions and identify exploration frontiers online, and integrates these estimates with a reactive controller for real-time navigation. Frontier selection is formulated as a multi-objective optimization that balances safe-set expansion probability and goal-directed cost, with subgoals selected via scalarization over the Pareto-optimal frontier set. PSANE safely explores unknown granular terrain and reaches specified goals using only proprioceptively estimated traversability, while achieving performance improvements over baseline methods.

Matthew Y. Jiang, Feifei Qian, Shipeng Liu

Georgia Institute of Technology, Atlanta, GA 30332, USA · University of Southern California, Los Angeles, CA 90089, USA

Planetary Rover Robotics Autonomous Robot Exploration Robotics Safe Robot Navigation Mobile Robotics Terrain-Aware Robot Locomotion

Ask PDF

May 23, 2026

Terrain-Adaptive Grouser Wheel for Optimal Planetary Exploration: Design and Experimental Investigation

Planetary rovers operating in extraterrestrial environments often encounter significant mobility challenges due to varying terrain features such as gradients and granularity. While recent works in multimodal wheel design have explored adjustments in stiffness, compliance, and diameter as a means to improve terrain adaptability, full wheel grouser-adjustable designs remain largely unexplored. Grousers are a compelling feature to actuate, as granular terrains tend to require increased grouser height for improved wheel performance. As a result, we introduce [Anonymized Robot Name], a multimodal wheel capable of continuously adjusting its grouser height for terrain adaptation. The platform was evaluated across four representative surfaces, including vinyl flooring, coarse rock, pea gravel, and sand under two packing states, spanning a range of granular conditions. Results from 750 experimental trials demonstrate that adaptive deployment reduces slip by 30.0--58.0% and improves travel time and energy consumption by up to 77.4% in granular regimes relative to fixed configurations. Using the terrain trial data, a simplified scaling analysis was developed and validated, suggesting a relationship between terrain granularity and optimal grouser height for the tested configuration. No single grouser height minimized slip across all terrains, underscoring the limitations of fixed-wheel systems commonly used for planetary exploration. This observation reinforces the potential of grouser-adaptive morphology, such as [Anonymized Robot Name], as an effective solution for enhancing rover mobility across diverse and mobility-challenging extraterrestrial environments.

Vincent Griffo, Yashwanth Kumar Nakka

Aerospace Robotics Lab, Daniel Guggenheim School of Aerospace Engineering, Georgia Institute of Technology, Atlanta, GA, USA

Granular Terrain Locomotion Robotics Robot Locomotion

Ask PDF

Abstract

Planetary Rover Robotics

Explore similar work

Learning All-Terrain Locomotion for a Planetary Rover with Actively Articulated Suspension

Proprioceptive Safe Active Navigation and Exploration for Planetary Environments

Terrain-Adaptive Grouser Wheel for Optimal Planetary Exploration: Design and Experimental Investigation