Improving Object Distance Estimation in Automated Driving Systems Using Camera Images, LiDAR Point Clouds and Hierarchical Clustering

As automated vehicles have been considered one of the important trends in intelligent transportation systems, various research is being conducted to enhance their safety. In particular, the importance of technologies for the design of preventive automated driving systems, such as detection of surrounding objects and estimation of distance between vehicles. Object detection is mainly performed through cameras and LiDAR, but due to the cost and limits of LiDAR’s recognition distance, the need to improve Camera recognition technique, which is relatively convenient for commercialization, is increasing. This study learned convolutional neural network (CNN)-based faster regions with CNN (Faster R-CNN) and You Only Look Once (YOLO) V2 to improve the recognition techniques of vehicle-mounted monocular cameras for the design of preventive automated driving systems, recognizing surrounding vehicles in black box highway driving videos and estimating distances from surrounding vehicles through more suitable models for automated driving systems. Moreover, we learned the PASCAL visual object classes (VOC) dataset for model comparison. Faster R-CNN showed similar accuracy, with a mean average precision (mAP) of 76.4 to YOLO with a mAP of 78.6, but with a Frame Per Second (FPS) of 5, showing slower processing speed than YOLO V2 with an FPS of 40, and a Faster R-CNN, which we had difficulty detecting. As a result, YOLO V2, which shows better performance in accuracy and processing speed, was determined to be a more suitable model for automated driving systems, further progressing in estimating the distance between vehicles. For distance estimation, we conducted coordinate value conversion through camera calibration and perspective transform, set the threshold to 0.7, and performed object detection and distance estimation, showing more than 80% accuracy for near-distance vehicles. Through this study, it is believed that it will be able to help prevent accidents in automated vehicles, and it is expected that additional research will provide various accident prevention alternatives such as calculating and securing appropriate safety distances, depending on the vehicle types.

Download Full-text

A Branch-Trunk-Constrained Hierarchical Clustering Method for Street Trees Individual Extraction from Mobile Laser Scanning Point Clouds

Measurement ◽

10.1016/j.measurement.2021.110440 ◽

2021 ◽

pp. 110440

Author(s):

Jintao Li ◽

Xiaojun Cheng ◽

Zhihua Xiao

Keyword(s):

Hierarchical Clustering ◽

Laser Scanning ◽

Point Clouds ◽

Clustering Method ◽

Street Trees

Download Full-text

Fast plane extraction in organized point clouds using agglomerative hierarchical clustering

2014 IEEE International Conference on Robotics and Automation (ICRA) ◽

10.1109/icra.2014.6907776 ◽

2014 ◽

Cited By ~ 62

Author(s):

Chen Feng ◽

Yuichi Taguchi ◽

Vineet R. Kamat

Keyword(s):

Hierarchical Clustering ◽

Point Clouds ◽

Agglomerative Hierarchical Clustering ◽

Plane Extraction

Download Full-text

Influence of Systematic Frequency-Sweep Non-Linearity on Object Distance Estimation in FMCW/FSCW Radar Systems

33rd European Microwave Conference, 2003 ◽

10.1109/euma.2003.340795 ◽

2003 ◽

Cited By ~ 16

Author(s):

Markus Pichler ◽

Andreas Stelzer ◽

Peter Gulden ◽

Martin Vossiek

Keyword(s):

Distance Estimation ◽

Frequency Sweep ◽

Object Distance ◽

Radar Systems

Download Full-text

Implementation of Stereo Vision Semi-Global Block Matching Methods for Distance Measurement

Indonesian Journal of Electrical Engineering and Computer Science ◽

10.11591/ijeecs.v12.i2.pp585-591 ◽

2018 ◽

Vol 12 (2) ◽

pp. 585 ◽

Cited By ~ 1

Author(s):

Raden Arief Setyawan ◽

Rudy Sunoko ◽

Mochammad Agus Choiron ◽

Panca Mudji Rahardjo

Keyword(s):

Stereo Vision ◽

Stereo Matching ◽

Distance Estimation ◽

Distance Measurement ◽

Block Matching ◽

Disparity Map ◽

Block Matching Algorithm ◽

Object Distance ◽

Left And Right ◽

3D Movie

Stereo vision has become an attractive topic research in the last decades. Many implementations such as the autonomous car, 3D movie, 3D object generation, are produced using this technique. The advantages of using two cameras in stereo vision are the disparity map between images. Disparity map will produce distance estimation of the object. Distance measurement is a crucial parameter for an autonomous car. The distance between corresponding points between the left and right images must be precisely measured to get an accurate distance. One of the most challenging in stereo vision is to find corresponding points between left and right images (stereo matching). This paper proposed distance measurement using stereo vision using Semi-Global Block Matching algorithm for stereo matching purpose. The object is captured using a calibrated stereo camera. The images pair then optimized using WLS Filter to reduce noises. The implementation results of this algorithm are furthermore converted to a metric unit for distance measurement. The result shows that the stereo vision distance measurement using Semi-Global Block Matching gives a good result. The obtained best result of this work contains error of less than 1% for 1m distance

Download Full-text

ESTIMATING AIRCRAFT HEADING BASED ON LASERSCANNER DERIVED POINT CLOUDS

ISPRS Annals of Photogrammetry Remote Sensing and Spatial Information Sciences ◽

10.5194/isprsannals-ii-3-w4-95-2015 ◽

2015 ◽

Vol II-3/W4 ◽

pp. 95-102 ◽

Cited By ~ 1

Author(s):

Z. Koppanyi ◽

C, K. Toth

Keyword(s):

Cross Sections ◽

Point Cloud ◽

Reference Model ◽

Laser Scanner ◽

Point Clouds ◽

Motion Direction ◽

Object Distance ◽

Aircraft Motion ◽

Heading Estimation ◽

Data Processing Methods

Using LiDAR sensors for tracking and monitoring an operating aircraft is a new application. In this paper, we present data processing methods to estimate the heading of a taxiing aircraft using laser point clouds. During the data acquisition, a Velodyne HDL-32E laser scanner tracked a moving Cessna 172 airplane. The point clouds captured at different times were used for heading estimation. After addressing the problem and specifying the equation of motion to reconstruct the aircraft point cloud from the consecutive scans, three methods are investigated here. The first requires a reference model to estimate the relative angle from the captured data by fitting different cross-sections (horizontal profiles). In the second approach, iterative closest point (ICP) method is used between the consecutive point clouds to determine the horizontal translation of the captured aircraft body. Regarding the ICP, three different versions were compared, namely, the ordinary 3D, 3-DoF 3D and 2-DoF 3D ICP. It was found that 2-DoF 3D ICP provides the best performance. Finally, the last algorithm searches for the unknown heading and velocity parameters by minimizing the volume of the reconstructed plane. The three methods were compared using three test datatypes which are distinguished by object-sensor distance, heading and velocity. We found that the ICP algorithm fails at long distances and when the aircraft motion direction perpendicular to the scan plane, but the first and the third methods give robust and accurate results at 40m object distance and at ~12 knots for a small Cessna airplane.

Download Full-text

PLASTIC SURGERY FOR 3D CITY MODELS: A PIPELINE FOR AUTOMATIC GEOMETRY REFINEMENT AND SEMANTIC ENRICHMENT

ISPRS Annals of Photogrammetry Remote Sensing and Spatial Information Sciences ◽

10.5194/isprs-annals-v-4-2021-17-2021 ◽

2021 ◽

Vol V-4-2021 ◽

pp. 17-24

Author(s):

O. Wysocki ◽

B. Schwab ◽

L. Hoegner ◽

T. H. Kolbe ◽

U. Stilla

Keyword(s):

Point Clouds ◽

3D Models ◽

Data Types ◽

Automated Driving ◽

3D City Models ◽

Semantic Enrichment ◽

Simulation Engine ◽

Global Accuracy ◽

Model Database ◽

City Models

Abstract. Nowadays, the number of connected devices providing unstructured data is rapidly rising. These devices acquire data with a temporal and spatial resolution at an unprecedented level creating an influx of geoinformation which, however, lacks semantic information. Simultaneously, structured datasets like semantic 3D city models are widely available and assure rich semantics and high global accuracy but are represented by rather coarse geometries. While the mentioned downsides curb the usability of these data types for nowadays’ applications, the fusion of both shall maximize their potential. Since testing and developing automated driving functions stands at the forefront of the challenges, we propose a pipeline fusing structured (CityGML and HD Map datasets) and unstructured datasets (MLS point clouds) to maximize their advantages in the automatic 3D road space models reconstruction domain. The pipeline is a parameterized end-to-end solution that integrates segmentation, reconstruction, and modeling tasks while ensuring geometric and semantic validity of models. Firstly, the segmentation of point clouds is supported by the transfer of semantics from a structured to an unstructured dataset. The distinction between horizontal- and vertical-like point cloud subsets enforces a further segmentation or an immediate refinement while only adequately depicted models by point clouds are allowed. Then, based on the classified and filtered point clouds the input 3D model geometries are refined. Building upon the refinement, the semantic enrichment of the 3D models is presented. The deployment of a simulation engine for automated driving research and a city model database tool underlines the versatility of possible application areas.

Download Full-text