End-to-End Monocular Range Estimation for Forward Collision Warning

Abstract

Estimating range to the closest object in front is the core component of the forward collision warning (FCW) system. Previous monocular range estimation methods mostly involve two sequential steps of object detection and range estimation. As a result, they are only effective for objects from specific categories relying on expensive object-level annotation for training, but not for unseen categories. In this paper, we present an end-to-end deep learning architecture to solve the above problems. Specifically, we represent the target range as a weighted sum of a set of potential distances. These potential distances are generated by inverse perspective projection based on intrinsic and extrinsic camera parameters, while a deep neural network predicts the corresponding weights of these distances. The whole architecture is optimized towards the range estimation task directly in an end-to-end manner with only the target range as supervision. As object category is not restricted in the training stage, the proposed method can generalize to objects with unseen categories. Furthermore, camera parameters are explicitly considered in the proposed method, making it able to generalize to images taken with different cameras and novel views. Additionally, the proposed method is not a pure black box, but provides partial interpretability by visualizing the produced weights to see which part of the image dominates the final result. We conduct experiments to verify the above properties of the proposed method on synthetic and real-world collected data.

Object Detection, Recognition, and Tracking Algorithms for ADASs—A Study on Recent Trends

“…[ 146 ] proposes a vision-based FCWS merging camera and radar data for real-time multi-vehicle detection, addressing challenging conditions like occlusions and lighting variations. Tang et al [ 147 ] introduce a monocular range estimation system using a single camera for precise FCWS, especially in difficult scenarios. Lim et al [ 148 ] suggest a smartphone-based FCWS for motorcyclists utilizing phone sensors to predict collision risks.…”

“…Lin, Dai, Wu, and Chen [ 146 ] introduce a ‘Driver Assistance System with Forward Collision and Overtaking Detection’. Tang and Li [ 147 ] propose ‘End-to-End Monocular Range Estimation’ for collision warning. Lim et al [ 148 ] created a ‘Forward Collision Warning System for Motorcyclists’ using smartphone sensors.…”

Probabilistic 3D motion model for object tracking in aerial applications

“…To alleviate inaccurate range estimations in different views, regression based on deep features can come in handy [62,[81][82][83][84][85]. Alike the FCN methods, the typical approach of deep regressors consists of two steps for object detection and distance regression.…”

“…While the previous methods were limited to a specific set of object classes that are detectable by object detectors, ref. [85] introduces a method to estimate the distance to unseen classes as well.…”

Sensors for Road Vehicles of the Future

“…In a specific case of the Forward Collision Warning assistance system, an end-to-end deep learning architecture to improve monocular range estimation methods is presented in [ 10 ]. In this case, the target range is represented as a weighted sum of a set of potential distances.…”