The Road Ahead in Autonomous Driving: The KITScenes Multimodal Dataset

2026-06-01

AI Summary

Existing autonomous driving datasets are limited in sensor fidelity, high-definition (HD) map completeness, and geographic diversity. This work proposes KITScenes Multimodal, a novel dataset that, for the first time in a publicly available benchmark, achieves reprojection-level accuracy in mapping full-element 3D traffic infrastructure. It integrates high-resolution global-shutter cameras, LiDAR with over 400-meter detection range, 4D imaging radar, and redundant GNSS/INS systems to enable synchronized multimodal data acquisition. Captured across irregular road networks and mixed-traffic scenarios in Europe, the dataset significantly enhances geographic diversity and map completeness. The authors introduce four core benchmark tasks—online HD mapping, long-range depth estimation, novel view synthesis, and end-to-end driving—and demonstrate the dataset’s effectiveness through evaluations in open-source frameworks.

Abstract

Existing autonomous driving datasets have enabled major progress, but fall short in sensor fidelity, map completeness, or geographic diversity. We present KITScenes Multimodal, a European dataset built around high-fidelity sensors and maps. Our fully synchronized sensor suite combines high-resolution global-shutter cameras, long-range lidar beyond 400m, 4D imaging radar, and redundant GNSS/INS localization. Our HD maps are, to our knowledge, the most complete of any sensor dataset, validated through autonomous driving trials on open-source software. For the first time in a public dataset, all driving-relevant traffic elements, such as traffic lights, are mapped in 3D to a reprojection-accurate level with full topological connectivity. Recorded in cities with irregular street layouts and mixed traffic modes, our dataset complements existing datasets by broadening the available geographic diversity. We also introduce four benchmarks, each advancing spatial learning for embodied AI: online HD map construction, long-range depth estimation, novel view synthesis, and end-to-end driving. Project page: https://kitscenes.com/

Problem

• autonomous driving
• sensor fidelity
• map completeness
• geographic diversity
• multimodal dataset

Innovation

• multimodal sensor fusion
• high-definition mapping
• 4D imaging radar
• geographic diversity
• reprojection-accurate 3D traffic elements