Robust Deep-Learning-Based Road-Prediction for Augmented Reality Navigation Systems

TL;DR

This paper introduces a deep learning-based method for robust road course prediction in augmented reality navigation, capable of functioning in adverse weather and without lane markings, by fusing camera, radar, and map data.

Contribution

It presents a multi-scale CNN trained on night-time images for accurate road pixel detection and a fusion framework for long-range road estimation in AR navigation.

Findings

Achieves state-of-the-art performance in road course estimation

Operates reliably without lane markings

Effective in adverse weather conditions

Abstract

This paper proposes an approach that predicts the road course from camera sensors leveraging deep learning techniques. Road pixels are identified by training a multi-scale convolutional neural network on a large number of full-scene-labeled night-time road images including adverse weather conditions. A framework is presented that applies the proposed approach to longer distance road course estimation, which is the basis for an augmented reality navigation application. In this framework long range sensor data (radar) and data from a map database are fused with short range sensor data (camera) to produce a precise longitudinal and lateral localization and road course estimation. The proposed approach reliably detects roads with and without lane markings and thus increases the robustness and availability of road course estimations and augmented reality navigation. Evaluations on an extensive set of high precision ground truth data taken from a differential GPS and an inertial measurement unit show that the proposed approach reaches state-of-the-art performance without the limitation of requiring existing lane markings.