Showing papers in "IEEE Intelligent Transportation Systems Magazine in 2017"

Autonomous Vehicle Safety: An Interdisciplinary Challenge

Abstract: Ensuring the safety of fully autonomous vehicles requires a multi-disciplinary approach across all the levels of functional hierarchy, from hardware fault tolerance, to resilient machine learning, to cooperating with humans driving conventional vehicles, to validating systems for operation in highly unstructured environments, to appropriate regulatory approaches. Significant open technical challenges include validating inductive learning in the face of novel environmental inputs and achieving the very high levels of dependability required for full-scale fleet deployment. However, the biggest challenge may be in creating an end-to-end design and deployment process that integrates the safety concerns of a myriad of technical specialties into a unified approach.

Dynamical Modeling and Distributed Control of Connected and Automated Vehicles: Challenges and Opportunities

Abstract: The platooning of connected and automated vehicles (CAVs) is expected to have a transformative impact on road transportation, e.g., enhancing highway safety, improving traffic utility, and reducing fuel consumption. Requiring only local information, distributed control schemes are scalable approaches to the coordination of multiple CAVs without using centralized communication and computation. From the perspective of multi-agent consensus control, this paper introduces a decomposition framework to model, analyze, and design the platoon system. In this framework, a platoon is naturally decomposed into four interrelated components, i.e., 1) node dynamics, 2) information flow network, 3) distributed controller, and 4) geometry formation. The classic model of each component is summarized according to the results of the literature survey; four main performance metrics, i.e., internal stability, stability margin, string stability, and coherence behavior, are discussed in the same fashion. Also, the basis of typical distributed control techniques is presented, including linear consensus control, distributed robust control, distributed sliding mode control, and distributed model predictive control.

Luxembourg SUMO Traffic (LuST) Scenario: Traffic Demand Evaluation

Abstract: Both the industrial and the scientific communities are working on problems related to vehicular traffic congestion, intelligent transportation systems, and mobility patterns using information collected from a variety of sources. Usually, a vehicular traffic simulator, with an appropriate scenario for the problem at hand, is used to reproduce realistic mobility patterns. Many mobility simulators are available, and the choice is made based on the type of simulation required, but a common problem is finding a realistic traffic scenario. The aim of this work is to provide and evaluate a scenario able to meet all the basic requirements in terms of size, realism, and duration, in order to have a common basis for evaluations. In the interest of building a realistic scenario, we used information from a real city with a typical topology common in mid-size European cities, and realistic traffic demand and mobility patterns. In this paper, we show the process used to build the Luxembourg SUMO Traffic (LuST) Scenario, and present a summary of its characteristics together with our evaluation and validation of the traffic demand and mobility patterns.

Scenario Model Predictive Control for Lane Change Assistance and Autonomous Driving on Highways

Abstract: This paper presents a novel design of control algorithms for lane change assistance and autonomous driving on highways, based on recent results in Scenario Model Predictive Control (SCMPC). The basic idea is to account for the uncertainty in the traffic environment by a small number of future scenarios, which is intuitive and computationally efficient. These scenarios can be generated by any model-based or data-based approach. The paper discusses the SCMPC design procedure, which is simple and can be generalized to other control challenges in automated driving, as well as the controller's robustness properties. Experimental results demonstrate the effectiveness of the SCMPC algorithm and its performance in lane change situations on highways.

Autonomous Vehicle-Intersection Coordination Method in a Connected Vehicle Environment

Abstract: In response to the need for developing coordinated schemes of autonomous vehicles (AVs) at an intersection. This paper presents a novel coordination method for intersection management in a connected vehicle environment. The road network is divided into three logical sections, namely, buffer area, core area and free driving area. In addition, a buffer-assignment mechanism is developed to cooperatively assign a specific crossing span for an individual AV and guide each AV to adjust its entry time and corresponding speed in the core area. A set-projection algorithm and a three-segment linear speed profile are employed to control the trajectories of the AVs in the buffer area. Furthermore, the assignment failure handling process and the crossing rule for human-driven vehicles are advanced to enhance the practicability and reliability of the buffer-assignment mechanism. The performance of the proposed method is evaluated by simulating various traffic conditions on an actual urban network. The simulation experiments and sensitivity analyses demonstrate that the proposed method can significantly reduce 24.2%-77.1% of travel delays, decrease almost 99% of number of stops and improve the sustainability of the traffic system by saving 22.1%-52% of fuel consumption.

Sensor Fusion Methodology for Vehicle Detection

Abstract: A novel sensor fusion methodology is presented, which provides intelligent vehicles with augmented environment information and knowledge, enabled by vision-based system, laser sensor and global positioning system. The presented approach achieves safer roads by data fusion techniques, especially in single-lane carriageways where casualties are higher than in other road classes, and focuses on the interplay between vehicle drivers and intelligent vehicles. The system is based on the reliability of laser scanner for obstacle detection, the use of camera based identification techniques and advanced tracking and data association algorithms i.e. Unscented Kalman Filter and Joint Probabilistic Data Association. The achieved results foster the implementation of the sensor fusion methodology in forthcoming Intelligent Transportation Systems.

Smart Parking Guidance, Monitoring and Reservations: A Review

Abstract: As the urban population is increasing, more and more cars are circulating in the city to search for parking spaces which contributes to the global problem of traffic congestion. To alleviate the parking problems, smart parking systems must be implemented. In this paper, the background on parking problems is introduced and relevant algorithms, systems, and techniques behind the smart parking are reviewed and discussed. This paper provides a good insight into the guidance, monitoring and reservations components of the smart car parking and directions to the future development.

Traffic Coordination at Road Intersections: Autonomous Decision-Making Algorithms Using Model-Based Heuristics

Abstract: This article focuses on the traffic coordination problem at traffic intersections. We present a decentralized coordination approach, combining optimal control with model-based heuristics. We show how model-based heuristics can lead to low-complexity solutions that are suitable for a fast online implementation, and analyze its properties in terms of efficiency, feasibility and optimality. Finally, simulation results for different scenarios are also presented.

Human Drivers Based Active-Passive Model for Automated Lane Change

Abstract: Lane change maneuver is a complicated maneuver, and incorrect maneuvering is an important reason for expressway accidents and fatalities. In this scenario, automated lane change has great potential to reduce the number of accidents. Previous research in this area, typically, focuses on the generation of an optimal lane change trajectory, while ignoring the human behavior model. To understand the human lane change behavior model, we carried out experiments on Japanese expressways. By analyzing the human-driver lane change data, we propose a two-segment lane change model that mimics the human-driver. We categorize the driving environment based on the observation grid and propose different lane change behaviors to handle the different scenarios. We develop an intuitive method to select the suitable lane change behavior, for a given scenario, using active (accelerate/decelerate) and passive (wait) information derived from the distance and related velocity (dx/dv) graph. Additionally, we also identify the most desirable and safe conditions for doing lane change based on the human driver preference data. We evaluated the proposed model by performing lane change simulations in the PreScan environment, while considering the vehicle motion/control model. The simulation results show the proposed model is able to handle complicated lane change scenarios with human driver-like performance.

Ready for Take-Over? A New Driver Assistance System for an Automated Classification of Driver Take-Over Readiness

Abstract: Recent studies analyzing driver behavior report that various factors may influence a driver's take-over readiness when resuming control after an automated driving section. However, there has been little effort made to transfer and integrate these findings into an automated system which classifies the driver's take-over readiness and derives the expected take-over quality. This study now introduces a new advanced driver assistance system to classify the driver's takeover readiness in conditionally automated driving scenarios. The proposed system works preemptively, i.e., the driver is warned in advance if a low take-over readiness is to be expected. The classification of the take-over readiness is based on three information sources: (i) the complexity of the traffic situation, (ii) the current secondary task of the driver, and (iii) the gazes at the road. An evaluation based on a driving simulator study with 81 subjects showed that the proposed system can detect the take-over readiness with an accuracy of 79%. Moreover, the impact of the character of the take-over intervention on the classification result is investigated. Finally, a proof of concept of the novel driver assistance system is provided showing that more than half of the drivers with a low take-over readiness would be warned preemptively with only a 13% false alarm rate.

Automatic Incident Detection in Intelligent Transportation Systems Using Aggregation of Traffic Parameters Collected Through V2I Communications

Abstract: Recent research in Intelligent Transportation System (ITS) focuses on Automatic Incident Detection (AID) techniques. Using advances in wireless networking and sensor technologies, modern vehicles have the ability to communicate with each other as well as with roadside infrastructure units (RSUs) in order to increase road safety. These new innovations in transportation technology provide traffic management system the ability to use data collected from the vehicles on the road to detect congestion and traffic incidents. Lately, many techniques were developed to alert drivers in advance about traffic incidents and to enable them to avoid congestion. In this paper, we discuss probabilistic collection of traffic data through vehicle-to-infrastructure (V2I) communications and present two novel techniques for automatic detection of traffic incidents in a highway scenario that are based on the use of distance and time for changing lanes, respectively vehicle speed changes over time. The proposed methods, which are illustrated with numerical results obtained from simulations, outperform alternative AID techniques through higher incident detection rates, about 25% shorter peak queue values and 20% faster dissipation of roadway congestion.

Implementation and Evaluation of a Cooperative Vehicle-to-Pedestrian Safety Application

Abstract: While the development of Vehicle-to-Vehicle (V2V) safety applications based on Dedicated Short-Range Communications (DSRC) has been extensively undergoing standardization for more than a decade, such applications are extremely missing for Vulnerable Road Users (VRUs). Nonexistence of collaborative systems between VRUs and vehicles was the main reason for this lack of attention. Recent developments in Wi-Fi Direct and DSRC-enabled smartphones are changing this perspective. Leveraging the existing V2V platforms, we propose a new framework using a DSRC-enabled smartphone to extend safety benefits to VRUs. The interoperability of applications between vehicles and portable DSRC enabled devices is achieved through the SAE J2735 Personal Safety Message (PSM). However, considering the fact that VRU movement dynamics, response times, and crash scenarios are fundamentally different from vehicles, a specific framework should be designed for VRU safety applications to study their performance. In this article, we first propose an end-to-end Vehicle-to-Pedestrian (V2P) framework to provide situational awareness and hazard detection based on the most common and injury-prone crash scenarios. The details of our VRU safety module, including target classification and collision detection algorithms, are explained next. Furthermore, we propose and evaluate a mitigating solution for congestion and power consumption issues in such systems. Finally, the whole system is implemented and analyzed for realistic crash scenarios.

Citywide Estimation of Traffic Dynamics via Sparse GPS Traces

Abstract: Traffic congestion is a perpetual challenge in metropolitan areas around the world. The ability to understand traffic dynamics is thus critical to effective traffic control and management. However, estimation of traffic conditions over a large-scale road network has proven to be a challenging task for two reasons: first, traffic conditions are intrinsically stochastic; second, the availability and quality of traffic data vary to a great extent. Traditional traffic monitoring systems that exist mostly on major roads and highways are insufficient to recover the traffic conditions for an entire network. Recent advances in GPS technology and the resulting rich data sets offer new opportunities to improve upon such traditional means, by providing much broader coverage of road networks. Despite that, such data are limited by their spatial-temporal sparsity in practice. To address these issues, we have developed a novel framework to estimate travel times, traversed paths, and missing values over a large-scale road network using sparse GPS traces. Our method consists of two phases. In the first phase, we adopt the shortest travel time criterion based on Wardrop's Principles in the map-matching process. With an improved traveltime allocation technique, we have achieved up to 52.5% relative error reduction in network travel times compared to a state-of-the-art method [1]. In the second phase, we estimate missing values using Compressed Sensing algorithm, thereby reducing the number of required measurements by 94.64%.

Intelligent Tires?A Review of Tire Characterization Literature

Abstract: In this article, previous literature on the intelligent tire has been reviewed, especially focusing on the estimation of tire characteristics such as tire forces and contact patch characteristics of a loaded rolling tire. The contact patch features and the global deformation of tire are identified as the key factors for the tire force estimation. In this review article, the measurement values to characterize these estimation parameters and the corresponding sensors are investigated. The estimated tire features and associated algorithms are presented based on the type of measurements. It is revealed that numerous dynamic features including wheel forces, slippage in contact, contact length on dry and wet roads, and hydroplaning can be identified by using the intelligent tire system. From this review, the authors of this article suggest that intelligent tire can be exploited to analyze the tire characteristics themselves besides the application for the advanced vehicle control. However, for the purpose of tire characteristic analysis, the well-established physical concepts must be employed into the estimation process. This review article is expected to provide resources to researchers or developers who are interested in the advanced vehicle controls integrated with intelligent tire system, and who want to utilize the intelligent tire for the purpose of tire performance analysis as well.

Lane-Based Saturation Degree Estimation for Signalized Intersections Using Travel Time Data

Abstract: Saturation degree estimation is a vital problem of signal timing optimization. However, classic loop-detector-based algorithms are not capable to capture the severity of oversaturation, since detectors are located in front of stop lines, and also cannot distinguish the saturated degree in different lane groups if detectors are located at an upstream position. In this paper, we present a new method to estimate the lane-based saturation degree using travel times. The method is simple and mainly depends on the parameters of signal cycles and the corresponding virtual cycles. The virtual cycle parameters are extracted by analyzing the data on travel times using the K-mean cluster analysis. Then, two models for the traffic demand saturated degree (TDSD) and the effectively used green time saturation degree (EUGTSD) are presented based on the traffic flow conservation during one signal cycle and the corresponding virtual cycle. The new method can overcome the defects of loop-detector-based algorithms, and it can be used to optimize the TDSD and the EUGTSD simultaneously. Finally, the precision of the two types of models is evaluated using field survey data. The results show that the new method has a higher precision for the TDSD and the same accuracy level for the EUGTSD compared to the existing methods. The findings of this paper have potential applicability to signal control systems.

Online Recognition of Driver-Activity Based on Visual Scanpath Classification

Abstract: The next step towards the fully automated vehicle is the level of conditional automation, where the automated driving system can take over the control and responsibility for a limited time interval. Nevertheless, take-over situations may occur, forcing the driver to resume the driving task. Despite such situations, the driver is able to perform secondary tasks during conditionally automated driving, hence a low take-over quality must be expected. Methods for Driver-Activity Recognition (DAR) usually extract features for the classification within a moving time window. In this paper, the first DAR architecture based on the driver's scanpath, which is extracted by means of dynamic clustering and symbolic aggregate approximation patterns, is introduced. To demonstrate the potential of this approach, it is compared to a state-of-the-art method based on the data of a driving simulator study with 82 subjects. The classification performance of both DAR approaches was examined for decreasing window sizes with regard to the recognition of different secondary tasks and the separability of drivers using a handheld or hands-free device. Compared to the state-of-the-art approach, the proposed method shows a classification accuracy increase of nearly 20%, a significant improvement of the overall classification performance, and is able to classify the secondary tasks of the driver even for short windows of a duration of 5 s, i.e. with little information.

On the Use of Monte-Carlo Simulation and Deep Fourier Neural Network in Lane Departure Warning

Abstract: To make improvements on vision-based lane departure warning systems (LDWS), a lane departure prediction (LDP) method based on Monte-Carlo simulation and deep Fourier neural network (DFNN) is proposed. Firstly, a closed-loop driver-vehicle-road (DVR) system model is built up and the parameters of the system, consisting of vehicle states, positioning and road conditions, are initialized by random sampling. After simulating a large number of DVR systems with random parameters, the obtained results are used as samples to train a DFNN which predicts the forthcoming maximum lateral deviation and is optimized by employing deep learning method. Then, a LDP strategy is proposed by combining the DFNN with a driver activity index, which takes driver adaptation into consideration. The experimental evaluation shows that the proposed lane departure warning algorithm can predict the lane departure event in time and reduce the false-warning rate of existing methods in a significant way. More importantly, the proposed technique enhances the system's functions of over-speed warning on curved road and over-steer warning on low-adhesion road.

Cooperative Localization with Reliable Confidence Domains Between Vehicles Sharing GNSS Pseudoranges Errors with No Base Station

Abstract: Cooperation between road vehicles through information exchange is a promising way to enhance their absolute and relative positions. This paper presents an approach for generating, sharing and applying Global Navigation Satellite System (GNSS) pseudorange corrections through a V2X communication network. Conventionally, differential corrections are generated by fixed base stations with known positions and sent to mobile users. Here, the proposed cooperative method has no central server and the estimation of the raw measurements errors is done in a fully distributed way. Using a model of the correlation of the pseudorange errors and through the knowledge of the local motions of the vehicles obtained by Dead Reckoning (DR) or tracking, a non linear observability shows that the estimation problem is solvable. A cooperative and fully distributed estimation method is then presented using Set Inversion and Constraint propagation techniques. Positions, pseudorange estimated errors and DR data are shared in the network of vehicles and confidence is handled by intervals, in a bounded error context. This allows computing highly reliable confidence domains with no direct range measurements, which is crucial for applications involving close proximity navigation. Indeed, the proposed data fusion framework does not require any linearization of the equations and is insensitive to the data incest problem since the same information can be exploited several times in the computation process without making the estimation over-converge. Results using real measurements are presented to illustrate the performance of the proposed cooperative method in comparison with standalone estimation. A classical sequential Bayesian method has also been implemented on the same data set and compared in terms of accuracy and confidence with a ground truth system.

On-Road Path Generation and Control for Waypoints Tracking

Abstract: This article addresses path generation and predictive control for waypoints tracking. By considering irregularity of waypoints given by the leading vehicle, the waypoints-following vehicle is made to track a clothoid path that assures safe and comfortable driving while keeping track of the waypoints in the least squares sense. In the generation of such a clothoid path, the path generation scheme utilized the clothoidal constraint and weighted least squares curve fitting. Predictive control is applied to improve path tracking performance over horizon, with constraints on steering and its rate being considered for safe and comfortable steering. A computational experiment, utilizing a set of leading vehicle generated waypoints on a highway, is conducted to demonstrate utility of the path generation and predictive control scheme.

Driving Characteristics and Adaptive Cruise Control ? A Naturalistic Driving Study

Abstract: With the increasing number of vehicles equipped with Adaptive Cruise Control (ACC), it becomes important to assess its impact on traffic flow efficiency, in particular with respect to capacity and queue discharge rate. Simulation studies and surveys suggest that ACC has both positive and negative effects on traffic flow, but empirical evidence on this topic is scarce. A naturalistic driving study has been conducted with 8 participants who drove their own ACC-equipped vehicle during their regular trips on freeways for a period of 4 to 5 weeks. We measured spacing, headway, speed, acceleration, lane use, and the number of lane changes, and compared these between ACC On and ACC Off in different traffic states, for a total of 48 hours of driving data. Results show that with ACC On, average spacing and headways were larger, whereas standard deviations were smaller. Larger headways can be assumed to reduce capacity, whereas more constant spacing, headway, speed, and acceleration indicate more stable traffic. With ACC On, drivers performed 36% fewer lane changes in saturated traffic, resulting in increased use of either the faster or the slower lane, depending on the driver. Furthermore we found that headways were smaller with ACC On than ACC Off when only selecting accelerations below -0.5m/s2 and above 0.5m/s2, which is the opposite of the overall finding. The latter result suggests that ACC has an important limitation: a lack of anticipation. On the other hand, the smaller headways with ACC On during acceleration indicate an increased queue discharge rate.

Design of Integrated Risk Management-Based Dynamic Driving Control of Automated Vehicles

Abstract: This paper describes the design of a fully automated driving algorithm for automated driving in complex urban scenarios and motorways with a satisfactory safety level. The proposed algorithm consists of the following three steps: surround recognition, motion planning, and vehicle control. The surround recognition system consists of three main modules: object classification, vehicle/non-vehicle tracking and dynamic drivable area determination. All system modules utilize information from potentially commercializable sensors such as vision sensors, radars, lidar and vehicle sensors. The objective of the motion planning module is to derive an optimal trajectory as a function of time and the surround recognition results. A dynamic drivable area is represented as a complete driving corridor that leads to the destination while making sure all objects are outside the left or right corridor bounds. In the case of moving objects such as other traffic participants, their behaviors are anticipated within the dynamic drivable area. The optimal trajectory planning uses the dynamic drivable area as a safety constraint and computes a trajectory in which the vehicle stays in its safe bounds considering the driver?s pattern and characteristics based on predicted risk potential. The developed algorithm has been evaluated by computer simulation and vehicle tests on urban roads and motorways.

Development of an Augmented Reality Vehicle for Driver Performance Evaluation

Abstract: Observing drivers' behaviors by reproducing traffic accidents and conflict situations is important for developing advanced driver assistance systems. For this purpose, driving simulators are frequently used to evaluate the effectiveness of driver assistance systems during product development. However, motion (simulator) sickness can be a serious practical problem with driving simulators. Therefore, an instrumented vehicle, the JARI-ARV (Japan Automobile Research Institute-Augmented Reality Vehicle), was developed to reproduce realistic traffic accident and conflict scenarios without endangering the driver. In this study, we examined level of control in the following cases: a right turn and encounter with a pedestrian by comparing the JARI-ARV with a standard (unaltered) same model vehicle. Results of the experiment indicated that drivers tend to react to virtual traffic participants in the same way as driving a standard vehicle. The study indicates that the JARI-ARV can play a useful role in human factors research.

Investigation of Route-Independent Aggressive and Safe Driving Features Obtained from Accelerometer Signals

Abstract: System that can identify safe and aggressive driving behavior is useful in nowadays transportation service sector: both in private and public. In this paper, we propose a systematic approach (step-by-step) for driving style classification and investigate features that best describe differences between aggressive and safe driving styles when driving information is collected from 3-axis accelerometer (G-sensor) alone which is ubiquitous and inexpensive inertial measurement sensor. Data were collected from light vehicle when driving in aggressive and save driving style. Features investigated do not depend on same route assumption and can be used as universal. Using driving features a classification task is performed by applying Random Forest algorithm. Results show that high classification accuracy (95.5%) using experimental data can be obtained when applying proposed input feature set.

Probabilistic Threat Assessment with Environment Description and Rule-based Multi-Traffic Prediction for Integrated Risk Management System

Abstract: The objective of this paper is to propose an original probabilistic threat assessment method to predict and avoid all possible kinds of collision in multi-vehicle traffics. The main concerns in risk assessment can be summarized as three requirements: 1) a description of a traffic situation containing the geometric description of the road, dynamic and static obstacle tracking, 2) a prediction of multiple traffics' reachable set under the reasonable behavior restriction, and 3) an assessment of collision risk which corresponds with driver sensitivity and can be applied to many complex situations without loss of generality. To fulfill these three requirements, the proposed algorithm for estimating the probability of collision occurrence of the ego vehicle follows the basic idea of the particle filtering and the collision probability can be numerically implemented and calculated. The overall performance of the proposed threat assessment algorithm is verified via vehicle tests in real road. It has been shown that the threat assessment performance for the given driving situations can be significantly enhanced by the proposed algorithm. And this enhancement of risk assessment performance led to capabilities improvement of driver assistance functions of ADASs.

Artificial Intelligence and Intelligent Transportation: Driving into the 3rd Axial Age with ITS

Abstract: Welcome to the inaugural issue of Artificial Intelligence Technologies for ITS section. I would like to open the section with a brief on the motivation for its introduction and the vision of its future.

Speed Tracking Based Energy-Efficient Freight Train Control Through Multi-Algorithms Combination

Abstract: Based on the characteristics of freight train control, which are nonlinear, time-delay, with multi-constraint and multiobjective, this paper focuses on speed tracking problem. Firstly, in a gradual process, a multi-modal fuzzy PID (MM-FPID) control algorithm is presented on the basis of a brief analysis of PI and PID control, which is generally used to train control in active services. Secondly, in order to deal with the time-delay problem of freight train, the paper adopts an approach of traction force feed-forward, which greatly improves the dynamic performance of the controller. Thirdly, for the overspeed brake problem caused by speed overshoot, the strategy of adaptive traction force limitation is adopted, and we get satisfactory results without increasing the safety speed margin. Fourthly, inspired by the selflearning characteristic of neural networks (NNs), an integrated controller of MM-FPID and NNs is proposed. Finally, with the help of a computer simulation platform, the paper puts forward a set of simulations, comparing the MM-FPID and the integrated control method with classical PID and fuzzy control. The results show that both MM-FPID and the integrated controller has satisfactory control effect, and their multi-modal structure makes it easy to fit different applications well, while the integrated controller has more potential in self-learning.

A Novel Approach for Traffic Signal Control: A Recommendation Perspective

Abstract: Traffic Signal Control is an effective mean of solving urban traffic problems by providing appropriate signal timings for road intersections. In this paper, we propose a novel approach originated from methods and applications of recommendation systems to select the most suitable signal timings for various traffic demands. Specifically, we use the idea of computational experiments and the well-known recommendation technology called Collaborative Filtering to find the best signal timings from a database filled with large amounts of traffic data. When the database is not large enough for some special traffic situations, traditional approaches are adopted. Corresponding data with good performance will be recorded.

Fast Branch Convolutional Neural Network for Traffic Sign Recognition

Abstract: In this paper, we propose a novel framework for speeding up the test-time of traffic sign recognition, which is named Branch Convolution Neural Network. It is the first time to introduce a branch-output mechanism into a deep Convolution Neural Network. Our model has an accuracy as high as a deep convolution neural network model, while it performs faster at the same condition during test stage. It is a significantly accelerated framework for designing a real-time deep neural network system. We present a detail process to change a regular pre-trained Convolution Neural Network into a Branch Convolution Neural Network: train several simple branch classifiers, bias classifiers and optimize branches. Experiment applied on GTSRB shows that large number of traffic signs are unnecessary to go through all layers in a deep model and they can be separated out in a relative shallow neural network. This framework speeds up the recognition progress, while keeping the accuracy within an extremely minor drop.

On the Similarities Between Urban Traffic Management and Communication Networks: Application of the Random Early Detection Algorithm for Self-Regulating Intersections

Abstract: Communications are revolutionizing transport systems. Recently, novel technologies from the fields of telecommunications and data networks have been incorporated into both vehicle systems and road infrastructures, setting new frontiers in the design and development of Smart Cities and Smart Urban Mobility. Whereas each scenario, i.e., the vehicular one and the data communications one, has its own and well-defined particularities, there are also similarities that should be further explored. In this work, we address this confluence of research areas by investigating the applicability in vehicle traffic management of an active queue control algorithm employed in communication networks and called Random Early Detection (RED). RED provides congestion avoidance in data networks by controlling the average queue size of network buffers. It detects incipient congestion and, using probabilistic packet-marking or packet-dropping behavior, it is able to keep the data network at high throughput and low delay. In this work, we propose an adaption of RED to vehicle traffic management in signaled intersections, namely, RED for Vehicles (REDV). A complete characterization of REDV is provided, including the impact of several configuration parameters. We also conduct an extensive comparative performance evaluation via computer simulation. Results show that REDV is a light and simple algorithm able to significantly reduce the average trip time, waiting time, and CO2 emissions in a signaled intersection compared to other approaches from the related literature.

Assistive Intelligent Transportation Systems: The Need for User Localization and Anonymous Disability Identification

Abstract: The main goal of Assistive Technology (AT) is to ensure the functional independence of disabled individuals. This paper proposes the definition of a new concept of AT within the context of the ITS, Assistive Intelligent Transportation System (AITS), analyzing its intrinsic requirements and providing a set of examples. We demonstrate that AITS must localize users with disabilities and identify their specific type of impairment in order to provide an efficient response, and we propose a specific procedure to guarantee anonymity while identifying the type of disability. Moreover, this new type of AT is illustrated by means of a new assistive intelligent pedestrian crossing application that is capable of localizing pedestrians with disabilities, identifying the specific type of impairment and providing an adaptive response to enhance functional capabilities of impaired pedestrians while crossing. By combining stereo-based object detection with radio-frequency identification technology (RFID and Bluetooth Low Energy), a specific solution to the problem of user localization and anonymous disability identification is proposed. Our approach has been validated in a real crosswalk scenario and it may be extended to other types of AITS, depending on the localization accuracy requirements and the range of operation of the specific application.