Guangquan Lu

A rear-end collision avoidance system of connected vehicles

The rear-end collision is one of the main types of accident, and it brings unnecessary casualties and property losses. Aiming at reducing the rear-end collision, some researchers focus on the front collision avoidance system. Generally, the rear-end collision avoidance system detects risk situation based on the information collected by radar or camera. When the following car is in risk of crash, the system will inform or warn the driver that the current speed is not safe, and help drivers to decelerate or brake. With the development of vehicle-to-vehicle communication (V2V), a new way to develop the rear-end collision avoidance system is put forward. By using the connected vehicles, the system can collect information and share message through the communicate equipment. In this paper, we introduce a collision avoidance system which is based on the vehicle-to-vehicle communication. This rear-end collision avoidance is with some driver behavior by using a car-following model based on risk perception. At last, we provide some experimental date about collision avoidance, and demonstrate the feasibility of this system in risk detection and crash avoidance.

A VANETs routing algorithm based on Euclidean distance clustering

Routing is a challenging task in the ad hoc networks, especially in vehicular ad hoc networks (VANETs) where the network topology changes fast and frequently. Since the nodes in VANETs are vehicles, which can easily provide the required power to run GPS receiver to get the accurate information of their position, the position-based routing is found to be a very promising routing strategy for VANETs. In this paper we present a clustering routing algorithm for VANETs. The clustering method is based on the Euclidean distance, which uses the position information to divide the vehicles into clusters. Furthermore, only the same direction vehicles can be divided into the same cluster. To reduce the flooding of the routing control message and increase the stability of the route, the routing discovery is also restricted by the vehicles driving direction. We implement the routing algorithm in NS2 and compare it with AODV, the simulation results show that in the same VANETs environment, the algorithm not only generate fewer routing control overhead, but also maintain stable route to transfmit more data packets.

A Rule Based Control Algorithm of Connected Vehicles in Uncontrolled Intersection

Aiming to address the safety issue for the uncontrolled intersection, the existing schemes including design optimization of the intersection structure and additional traffic signal layout will waste lots of resources. With the rapid development of intelligent transportation, the technology of vehicle-vehicle communication provides a new way for this problem. The paper proposed a set of rules to clarify the sequence of vehicles to pass through uncontrolled intersection. The rules are planned based on the law of road traffic safety. According to the rules, each approaching car makes decision for preempting or yielding other cars based on the information from vehicle-vehicle communication. If the approaching car needs to yield other cars, we propose an algorithm to find a proper deceleration value to do yielding. The car brakes automatically using this deceleration value to avoid collision with other cars. After all, the tests were done to verify the effectiveness of the algorithm and demonstrate the function among the connected vehicles in intersection. The rule based collision avoidance algorithm can provide real-time collision detection and make safe deceleration for the cars crossing the uncontrolled intersection.

A Bayesian Compressive Sensing Vehicular Location Method Based on Three-Dimensional Radio Frequency

In vehicular ad hoc networks (VANETs) safety applications, vehicular position is fundamental information to achieve collision avoidance and fleet management. Now, position information is comprehensively provided by global positioning system (GPS). However, in the dense urban, due to multipath effect and signal occlusion, GPS-based positioning method potentially fails to provide accurate position information. For this reason, an assistant approach has been presented in this paper by using three-dimensional radio frequency, such as time of arrival (TOA) and direction of arrival (DOA). With the goal of providing an efficient and reliable estimation of vehicular position in general traffic scenarios, we propose a hybrid TOA/DOA positioning method based on Bayesian compressive sensing (BCS), which benefits from the realization of vehicle-to-roadside wireless interaction with the dedicated short range communication. The effectiveness of the proposed approach is proved through extensive experiments in several scenarios where different signal configurations and the noise conditions are taken into account. Moreover, some comparative experiments are also performed to confirm the strength of our proposed approach.

A vehicular ad hoc networks intrusion detection system based on BUSNet

The open medium, dynamic topology, and multi-hop cooperative routing of vehicular ad hoc networks (VANETs) make it facing more security challenge than wired networks. In this paper, a hierarchical VANETs intrusion detection system based on BUSNet is present. BUSNet is basically a virtual mobile backbone infrastructure that is constructed using public buses. We use the bus nodes as the cluster-heads to gather the routing control messages and data packets transmitted among the vehicles. The bus nodes first transmit the original network behavior information to the access points deployed along the road sides. Then the access points can get a global view of the VANETs, and we can detect anomaly behaviors through analyzing the data. The anomaly detection method is based on the neural network which can build the normal network behavior model through learning process. After the trained neural network is stable, it can monitor the VANETs security by detecting the network control message and data packet in real time and alarm immediately if there is anomaly behavior. The experiments in NS2 show that the detection method can detect anomaly behavior with low false alarm rate.

Real-Time Vehicle Route Guidance Based on Connected Vehicles

With advances in connected vehicle technology, real-time vehicle route guidance systems gradually become indispensable equipments for drivers. Conventional route guidance systems are designed to direct a vehicle along the shortest path from the origin to the destination without considering the dynamic traffic information. Therefore the state-of-the-art route guidance systems incorporate real-time traffic information to find better paths. So, this paper presents a novel approach to realize the real-time vehicle route guidance. It focuses on the way to determine the optimal route based on the dynamic road segments division and the traditional Dijkstra algorithm. This approach divides the road to sub-sections with the traffic information, so the state of each segment can be easily figured out (i.e. average speed or average travel time). Then a dynamic road network graph can be drawn out. And traditional Dijkstra algorithm can find optimal route on it. A simulation is implemented to show optimal route at different time. It also compares with the traditional Dijkstra algorithm and the result is validated.

Travel time reliability affected by accident in freeway with connected vehicles

Road accidents usually cause great congestions, and shock waves generate due to the congestions, which severely affect travel time reliability (TTR) of upstream vehicles. Connected vehicles can get and spread information of the accident vehicles, which assists drivers in estimating TTR of the road and making better route choice. The major work of this paper is the proposal of a TTR model to calculate the TTR of freeway affected by accident. Information of time delay, accident position, accident duration and traffic flow status, which can be obtained from connected vehicles, are used. To better explain the TTR model, a simulation case analysis is given, effect of accident duration variance on TTR is also tested. Results show that TTR model in freeway affected by accident based on connected vehicles is practical, and the TTR increases with the variance. The research results are helpful to drivers.

A Road Safety Evaluation Method Based on Clustering Neural Network

Traffic accident records data mining is very important to understand why traffic accidents occurred frequently under some driving, environment, and vehicle conditions. There are many reasons can lead to accident, and their relationships are complex, it is very difficult to build a correct evaluation model. To overcome this problem, statistical models such as neural network, fuzzy logic, decision tree etc. have been widely used on such accident data to analyze road crashes. In this paper we present a road safety evaluation method based on the clustering neural network. This method first learn the history data, after it is stable, it can be used to evaluate the road safety. The experimental results prove that this method is effective.

Rollover warning algorithm based-on velocity for buses

This paper presents a rollover warning algorithm based-on velocity for buses, which can assist bus driver to resist rollover around the curve, especially around off-ramp. The paper mainly focused on a dynamic roll stability analysis and a warning method based-on velocity. At first dynamic roll stability limit (RSL) was designed for analyzing dynamic roll stability. And then a warning velocity for alerting drivers to slow down is determined using the measurement of lateral acceleration and the vehicle turning radius. The proposed rollover warning algorithm is evaluated by TruckSim software and simulation results show that the proposed warning velocity can give appropriate prediction of vehicle rollover in the typical scenario.