Robot-Centric Elevation Map Completion with Sensor Geometry-Aware Augmentation and Uncertainty Estimation
Open Access
- 19 August 2026
- journal article
- research article
- Published by MDPI AG in Applied Sciences
- Vol. 16 (16) , 8262
Abstract
Robot-centric elevation maps built from onboard sensing are always incomplete: occlusions, a limited field of view, and range limits leave large unobserved regions that traversability analysis and motion planning must still reason about. We present a supervised framework that completes these maps and reports a per-cell uncertainty. Its core is a ray-cone augmentation that removes angular sectors anchored at the sensor origin during training; unlike the random masks of image inpainting, these sectors match the coverage gaps of real deployments, such as camera failures or reduced camera configurations. Partial maps generated from four depth cameras along legged-robot trajectories in the TartanGround dataset are paired with dense ground truth, yielding 32,329 samples across five outdoor environments. An encoder–decoder network is trained with a masked -NLL loss and evaluated with a five-fold leave-one-environment-out protocol. The augmentation lowers the completion error on missing sensor sectors by 8.3 to 9.7%, depending on the sector width, at no measurable cost on uncorrupted partial inputs. The completed maps reach a hole root-mean-square error of 2.86 m, a 45% improvement over the strongest classical interpolation baseline.
Keywords
Funding Information
- EU NextGenerationEU (09I05-03-V02-00039)
This publication has 28 references indexed in Scilit:
- Generation of vehicle-free digital elevation models in open-pit mines using UAV image inpaintingMeasurement, 2025
- On the Uncertainty of Self-Supervised Monocular Depth Estimation2020
- Image Inpainting for Irregular Holes Using Partial Convolutions2018
- Uncertainty Estimates and Multi-hypotheses Networks for Optical Flow2018
- Concurrent Spatial and Channel ‘Squeeze & Excitation’ in Fully Convolutional Networks2018
- Enhanced Deep Residual Networks for Single Image Super-Resolution2017
- Deep Residual Learning for Image Recognition2016
- U-Net: Convolutional Networks for Biomedical Image Segmentation2015
- Navier-stokes, fluid dynamics, and image and video inpainting2005
- An Image Inpainting Technique Based on the Fast Marching MethodJournal of Graphics Tools, 2004