cs.ROOct 22, 2026

PATH: Preference-Aware Traversability Estimation for Robotic Teleoperation in Off-Road Navigation

Authors: Xianghao Meng, Wenshuo Wang, Cheng Shao, Lewei Feng, Qingxiao Liu, Canyu Wang, Junqiang Xi

Organizations: School of Mechanical Engineering, Beijing Institute of Technology, Beijing, China, 100081. · Hong Kong Polytechnic University, 11 Yuk Choi Road, Hung Hom, Kowloon, Hong Kong, 999077. · China North Vehicle Research Institute, Bejing, China, 100072.

Read PDF
Ask AI
Cite

Copy citation

DOINot available for this paper

Abstract

Traversability estimation is critical for ground vehicle teleoperation, yet existing methods often neglect the human operator's preferences for speed and risk tolerance. To address this gap, we propose PATH-Net, a Preference-Aware Traversability estimation Hierarchical Network that adaptively generates preference-conditioned traversability cost maps and speed profiles for off-road teleoperation. Our framework integrates visual and geometric bird's-eye-view features with short-term historical vehicle states through a multi-head shared-backbone architecture. Our PATH-Net infers traversability cost maps while optimizing speed recommendations by using maximum entropy inverse reinforcement learning and sparse speed supervision. Our method incorporates operator preference via an asymmetric feature aggregation mechanism, splitting inference into a cost branch that prioritizes high-risk traversability regions and a speed branch that outputs preference-adjusted recommended speeds. Experiments with 15 operators across 225 teleoperated trajectories show alignment with user-specific navigation strategies, achieving modified Hausdorff distance of 0.17 m (aggressive) and 0.15 m (conservative). In the questionnaire assessment, participants reported lower NASA-TLX workload scores with PATH-Net recommendations than without recommendations, decreasing from 34.7 to 22.2. These results support preference-conditioned trajectory and speed recommendation for off-road teleoperation.

Figures & tables

Fig. 1: Our PATH-Net integrates environmental factors, kinematics, and expert preferences to provide off-road traversability recommendations.
Fig. 2: Overview of PATH-Net. Visual-geometric BEV features and recent vehicle states history are fused to predict candidate cost and speed maps, which are then converted into preference-conditioned cost and speed maps by upper-tail cost aggregation and quantile-based speed selection.
Fig. 3: The spatiotemporal feature fusion module. The geometric guidance map is fused with visual features to form the environment feature map, while the kinematic history is encoded by a TCN and spatially tiled. Their concatenation yields the fused feature map used for subsequent cost and speed prediction.
Preference CategoryAssigned ρ\rhoAggregation Level α\alphaParticipants
Aggressive−0.9-0.90.050.054
Neutral0.00.00.500.506
Conservative+0.9+0.90.950.955
Total--15
TABLE I: Questionnaire-Defined Preferences and Aggregation Parameters
Fig. 4: Experimental hardware and teleoperation system.
Fig. 5: Training convergence and validation-trajectory alignment. (a) Training objectives of PATH-Net over 50 epochs, including the total objective L\mathcal{L} , IRL loss LIRL\mathcal{L}_{\mathrm{IRL}} , and speed regression loss Lspeed\mathcal{L}_{\mathrm{speed}} . (b) Validation pMHD of PATH-Net and the adapted baselines under the same 50-epoch training budget.
Fig. 6: Qualitative comparison of cost-related and speed-related outputs for a representative held-out off-road test case. The second row shows the final costmaps, the third row shows the corresponding speed-related outputs. PATH-Net and Velociraptor generate spatial speedmaps, whereas HDIF generates a costmap using a fixed speed query, and its speed-related visualization is derived from the common planner rather than a learned speedmap.
MethodMetricAggressiveConservative
VelociraptorpMHD (m)0.37±0.080.37\pm 0.080.36±0.070.36\pm 0.07
sMHD (m/s)0.23±0.060.23\pm 0.060.21±0.050.21\pm 0.05
HDIFpMHD (m)0.39±0.090.39\pm 0.090.38±0.080.38\pm 0.08
sMHD (m/s)––
PATH-Net, neutralpMHD (m)0.31±0.060.31\pm 0.060.29±0.060.29\pm 0.06
sMHD (m/s)0.18±0.040.18\pm 0.040.17±0.040.17\pm 0.04
TABLE II: Trajectory Alignment with Held-out Demonstrations (mean ±\pm SD over Test Trajectories; lower is better).
Fig. 7: Dataset-level operating points comparison. Human demonstrations are shown by their executed speed and path-averaged cost. Each PATH-Net marker shows the mean recommended speed and mean path-averaged cost obtained by applying one preference setting to the same recorded cases. The curve indicates the non-dominated frontier of the human demonstrations.
Workload SubscaleNo RecommendationWith PATH-NetDecrease (%)
Mental Demand42.3\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 21.4)}31.3\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 17.5)}26.0
Physical Demand24.1\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 17.9)}15.6\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 13.4)}35.2
Temporal Demand34.3\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 20.4)}19.5\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 11.7)}43.2
Performance Burden30.1\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 22.7)}25.1\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 22.5)}16.6
Effort46.4\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 20.5)}30.1\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 18.2)}35.2
Frustration31.0\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 15.0)}11.6\,{\color[rgb]{0.45,0.45,0.45}\scriptstyle(\pm 12.3)}62.6
TABLE III: NASA-TLX Workload Ratings under Different Recommendation Conditions. SD values are shown in gray parentheses.
Preference Setting ( ρ\rho )Min. Clearance (m)Avg. Recommended Speed (m/s)Path-Averaged Cost
Aggressive ( −0.9-0.9 )0.180.181.071.070.360.36
Neutral ( 0.00.0 )0.350.350.680.680.220.22
Conservative ( +0.9+0.9 )0.620.620.430.430.140.14
TABLE IV: Recommended Trajectory Measures under Different Preference Settings. Values are averaged over held-out test cases.
Fig. 8: Examples of PATH-Net recommended trajectories under different preference settings with the historic trajectory (white curve), held-out demonstration (red curve), and PATH-Net recommended trajectory (black curve).
Fig. 9: Aggregated costmaps and speedmaps under different preference settings. All costmaps use the range [0,1][0,1] , and all speedmaps share the range [0,vmax⁡][0,v_{\max}] . Moving from the aggressive to the conservative setting increases the contribution of larger candidate costs and lowers the recommended speed bound.
Figure 14
Figure 15
Figure 16
Figure 17
Figure 18
Figure 19
Figure 20
More in this topic

Robot Traversability Estimation

See all 28 papers

Explore similar work

CardsList

  1. Tele-Traversability: Rethinking Traversability for Teleoperated Ground Robots in Terrain Navigation

    Sep 17, 2026Lewei Feng, Qi Chen, Wenshuo Wang +6Embodied NavigationRobot Navigation

    School of Mechanical Engineering, Beijing Institute of Technology, China · Faculty of Marine Science and Technology, Beijing Institute of Technology, China · Department of Civil and Environmental Engineering, University of Auckland, New Zealand +2

  2. Which Terrain Is Better? Preference Learning with VLM Prototypes for Off-Road Traversability Ranking

    Sep 20, 2026Ji-Hoon Hwang, Jisung Bae, E-In Son +3Visual NavigationOff-Road Robot Navigation

  3. ProTrav: Probabilistic Traversability with Elevation Uncertainty Propagation for Risk-aware Off-road Navigation

    Oct 22, 2026Chuanlin Zuo, Miloš Prágr, Arun Kumar Singh +2Robot Traversability EstimationOff-Road Robot Navigation

    Faculty of Engineering and Natural Sciences, Tampere University, Finland. · Institute of Technology, University of Tartu, Estonia. · AI and ML Department, Volvo Autonomous Solutions, Sweden.