OORD: The Oxford Offroad Radar Dataset

TL;DR

OORD introduces the Oxford Offroad Radar Dataset, a large-scale, weather-robust off-road radar collection with GPS/INS ground truth across four rugged routes in the Scottish Highlands. It provides a comprehensive benchmark for radar place recognition, including open-source neural-network weights and a software toolkit to evaluate both radar-specific and pretrained baselines. The work demonstrates radar’s resilience to appearance changes and adverse conditions compared to visual modalities, while offering detailed calibration and data-access tooling to support reproducibility and rapid experimentation. By enabling cross-route relocalisation in natural environments, OORD aims to accelerate radar-centric SLAM, localisation, and sensor-fusion research in challenging outdoor settings.

Abstract

There is a growing academic interest as well as commercial exploitation of millimetre-wave scanning radar for autonomous vehicle localisation and scene understanding. Although several datasets to support this research area have been released, they are primarily focused on urban or semi-urban environments. Nevertheless, rugged offroad deployments are important application areas which also present unique challenges and opportunities for this sensor technology. Therefore, the Oxford Offroad Radar Dataset (OORD) presents data collected in the rugged Scottish highlands in extreme weather. The radar data we offer to the community are accompanied by GPS/INS reference - to further stimulate research in radar place recognition. In total we release over 90GiB of radar scans as well as GPS and IMU readings by driving a diverse set of four routes over 11 forays, totalling approximately 154km of rugged driving. This is an area increasingly explored in literature, and we therefore present and release examples of recent open-sourced radar place recognition systems and their performance on our dataset. This includes a learned neural network, the weights of which we also release. The data and tools are made freely available to the community at https://oxford-robotics-institute.github.io/oord-dataset.

Figures (8)

• Figure 1: Our data collection site is in the Scottish Highlands, with comprehensive coverage of Ardverikie Estate, close to the historic boundary between Lochaber and Badenoch. This image is taken on Lochan na h-Earba from our \ref{['DA1']} and \ref{['DA2']}4.4.2 datasets (\ref{['sec:twolochs', 'sec:routes_summary']}), showing unpaved terrain over, uneven landscape next to the vehicle, and the inclement weather. Images, GPS traces, and example radar scans for more specific areas of Ardverikie Estate are provided in \ref{['fig:dataset_overview_a', 'fig:dataset_overview_b']}.
• Figure 2: The bellmouth and hydro datasets, with GPS traces, trajectory-to-trajectory ground truth GPS matrices, and some sample radar scans.
• Figure 3: 4.1The maree and twolochs datasets, with GPS traces, ground truth matrices, and sample scans.
• Figure 4: Changes in elevation (measured as UTM down) for example forays from each challenge site. 4.3The horizontal axis shows time elapsed while driving (s) and the vertical axis shows UTM down (m) as a measure of elevation.
• Figure 5: 4.1.3On-board camera images for each released foray in \ref{['sec:routes_summary']}. Note that as this dataset is radar-focused, we do not release these camera images, but they are included here to give a sense of the inclement collection conditions.