Innovations for Smart and Connected Traffic. Guest Editor.
Professor Zhibin Li, Southeast
University, China
Driver’s black box: a system for driver
risk assessment using machine learning
and fuzzy logic
A. S. Yukse
Pages 482-500 | Received 25 Sep 2019, Accepted 13 Nov 2020, Published online: 02 Dec 2020
Abstract
Risky driving behaviors can cause accidents, which may result in major
material and moral damages. Due to the increase in road accidents, it has
become an important issue to identify risky driving behaviors and reward
people who drive safely. With the development of technology, it is now
possible to model driving behavior through advanced sensors integrated
into embedded systems. In this study, we modeled four major risky
driving behaviors and created driver profiles using data obtained from
accelerometer and gyroscope sensors and applying widely used machine
learning algorithms in behavior analysis, including the C4.5 Decision Tree,
Random Forest, Artificial Neural Network, Support-Vector Machine, K-
Nearest Neighbor, Naive Bayes, and K-Star algorithms. Risky driving
behaviors and their risk levels were evaluated in accordance with the
expert opinions of traffic officers, and driver risk was modeled using the
fuzzy logic method. The applied machine learning algorithms were
compared using common validation metrics such as accuracy, f-measure,
precision, and recall. In our experiments, the K-Star algorithm was the
most successful algorithm, with 100% accuracy. As a result, a highly
accurate, low-cost system which acts as the driver’s black box was
developed. The system can be integrated into vehicles and it can record
the driver’s behaviors and identify the risky ones. It can also open up new
horizons for insurance companies to utilize usage-based policies, in which
customers who drive safely are rewarded with lower car insurance
premiums, encouraging others to do the same.
Innovations for Smart and Connected Traffic. Guest Editor. Professor Zhibin Li, Southeast
University, China
Estimation of lane-level travel time
distributions under a connected
environment
Lili Lu Zhengbing He
Pages 501-512 | Received 21 Aug 2019, Accepted 18 Nov 2020, Published online: 03 Feb 2021
Abstract
Travel time distribution estimation is fundamentally important for the
evaluation of travel time variability and reliability. For urban roads, signal
delays are key components of travel time. They are stochastic and differ
for vehicular movements due to different signal timings for through, left-
turning, and right-turning vehicles. To better assist travelers in making trip
decisions under connected environments, this study seeks to investigate
lane-level travel time distributions for signalized arterial roads by
specifically considering the impacts of both the link travel time and the
vehicle movement-based signal delays at an intersection. A simulation
testbed based on the VISSIM microscopic traffic simulator and a Java
plugin is developed to mimic the traffic flow dynamics of a signalized
arterial, El Camino Real, in Palo Alto, California, under a connected
environment. The detailed vehicle trajectory data obtained from the
simulation can be leveraged to obtain lane-level travel time information.
Typical normal, lognormal, gamma, and Weibull distributions, as well as
kernel density estimation (KDE), are adopted to calibrate the lane-level
travel time distributions. The estimation results demonstrate that
conventional distribution models are suitable for estimating the travel
time distributions of only a few road segments, while KDE fully captures
travel time reliability metrics such as the buffer time index, skewness, and
width of the travel time distributions for all road segments. This result will
help traffic managers and engineers carry out effective traffic
management and control to optimize the operation of arterial roads.
Innovations for Smart and Connected Traffic. Guest Editor. Professor Zhibin Li, Southeast
University, China
Multilevel weather detection based on
images: a machine learning approach
with histogram of oriented gradient and
local binary pattern-based features
Md Nasim Khan
Pages 513-532 | Received 30 Dec 2019, Accepted 13 Jun 2021, Published online: 05 Jul 2021
Abstract
The primary objective of this study was to develop a trajectory-level
weather detection system capable of providing real-time weather
information at the road surface level using only a single video camera.
Two texture-based features, including histogram of oriented gradient
(HOG) and local binary pattern (LBP), were extracted from images and
used as classification parameters to train the weather detection models
using several machine learning classifiers, such as gradient boosting (GB),
random forest (RF), and support vector machine (SVM). In addition, a
unique multilevel model, based on a hierarchical structure, was also
proposed to increase detection accuracy. Evaluation results revealed that
the multilevel model provided an overall accuracy of 89.2%, which is 3.2%,
7.5%, and 7.9% higher compared to the SVM, RF, and GB model,
respectively, using the HOG features. Considering the LBP features, the
multilevel model also produced the best performance with an overall
accuracy of 91%, which is 1.6%, 8.6%, and 9% higher compared to the
SVM, RF, and GB models, respectively. A sensitivity analysis using the
proposed multilevel model revealed that the classification accuracy
improved with the increasing number of HOG and LBP features at the
expense of more computational powers. The proposed weather detection
method is cost-efficient and can be made widely available mainly due to
the recent booming of smartphone cameras and can be used to expand
and update the current weather-based variable speed limit (VSL) systems
in a connected vehicle (CV) environment.
Original Articles
A computer vision algorithm for locating
and recognizing traffic signal control light
status and countdown time
Xiaofeng Chen Yu Chen & Guohui Zhang
Pages 533-546 | Received 20 May 2018, Accepted 31 Dec 2020, Published online: 14 Jan 2021
Abstract
It is of practical importance for individual vehicles to automatically identify
traffic signal light status to facilitate their decision makings for traffic
safety performance enhancement and operation efficiency maximization
when they are approaching intersections, especially in near-future fully-or
semi-autonomous vehicle-penetrated traffic systems. Traditional methods
aim to directly extract information from the light emitting area of a traffic
signal light head, but this can easily result in missed or false signal light
status identification due to changes in detection distance and lighting
conditions. To address these research gaps, an innovative computer
vision-based algorithm is developed for signal control light status
identification and countdown time recognition. Based on the unique
characteristics of black traffic light boards, color segmentation is
conducted in both RGB and HSV channels, and traffic light boards and
countdown time display boards are located through cascade filtering
algorithms. In the HSV color space, the H component can be extracted and
highlighted to determine the current signal status by distinguishing the
distribution of the pixel points of the H component. Based on the nearest
neighbor interpolation algorithm, an adaptive threading method is
developed to improve countdown time recognition. The experimental
tests were conducted to verify the effectiveness of the developed
approach under various scenarios with acceptable recognition accuracy
and execution efficiency. The developed approach can be applied and
integrated with other detection and control algorithms to strengthen
vehicle-crossing safety performance and operational efficiency at
intersections.
Original Articles
Variable speed limit strategy with
anticipatory lane changing decisions
Bidoura Khondaker & Lina Kattan
Pages 547-559 | Received 03 May 2018, Accepted 09 Feb 2021, Published online: 07 Mar 2021
Abstract
This paper develops a novel anticipatory Variable Speed Limit (VSL) control
strategy that incorporates driver behavior based on trajectory data from
probe vehicles. In particular, the research examines how the speed limit
controls can be coordinated and optimized to reduce lane changing and
overall braking, thus to achieve greater traffic throughput, safety, and
sustainability. Driver behavior, such as acceleration/deceleration and lane
changing, as reflected in individual trajectory data from GPS enabled
probe vehicles are crucial for early detection of shockwave formation and
proactive selection of speed limits; and consequently delay, or even
eliminate breakdown formation. The core of this approach is the
incorporation of a lane changing model which provides a more robust
integrated speed limit selection and shockwave detection framework than
the one we have developed in an earlier paper. The control principle is
formulated in a generic fashion that finds the optimal speed limit control
variables for either separate or simultaneous reduction of travel time,
crash rates, and fuel consumption over a prediction horizon. The findings
from the paper suggested that the VSL strategy was able to result in fewer
lane changing rate (LCR) compared to the No-VSL case. Also, the
developed algorithm was able to produce higher frequency of lower
acceleration and deceleration rates than the No-VSL case, which indicated
a smoother acceleration and deceleration pattern that corresponded to
lower emission and fuel consumption. The performance of the algorithm
was also examined under different probe penetration rates and
congestion levels to identify the % of probe needed to achieve
simultaneous mobility, safety, and environmental objectives.