Manolis Tsogas

Sensor Fusion for Predicting Vehicles' Path for Collision Avoidance Systems

Path prediction is the only way that an active safety system can predict a drivers intention. In this paper, a model-based description of the traffic environment is presented - both vehicles and infrastructure - in order to provide, in real time, sufficient information for an accurate prediction of the ego-vehicles path. The proposed approach is a hierarchical-structured algorithm that fuses traffic environment data with car dynamics in order to accurately predict the trajectory of the ego-vehicle, allowing the active safety system to inform, warn the driver, or intervene when critical situations occur. The algorithms are tested with real data, under normal conditions, for collision warning (CW) and vision-enhancement applications. The results clearly show that this approach allows a dynamic situation and threat assessment and can enhance the capabilities of adaptive cruise control and CW functions by reducing the false alarm rate.

A Situation-Adaptive Lane-Keeping Support System: Overview of the SAFELANE Approach

Going beyond standard lane-departure-avoidance systems, this paper addresses the development of a system that is able to deal with a large set of different traffic situations. Its foundation lies on a thoroughly constituted environment detection through which a decision system is built. From the output of the decision module, the driver is warned or corrected through suited actuators that are coupled to control strategies. The input to the system comes from cameras, which are supplemented by active sensors (such as radar and laser scanners) and vehicle dynamic data, digital road maps, and precise vehicle-positioning data. In this paper, the presented system design is divided into three layers: the perception layer, which is responsible for the environment perception, and the decision and action layers, which are responsible for evaluating and executing actions, respectively.

An Advanced Cooperative Path Prediction Algorithm for Safety Applications in Vehicular Networks

Vehicular ad hoc networks (VANETs) are in the heart of current and future automotive research. Most of the current vehicular safety applications are based on sensors installed on the vehicle, e.g., radars and laserscanners. Due to the evolution of wireless networks, there is a tendency to exploit the cooperation among vehicles to enhance road safety through the related applications. Path prediction of a drivers own vehicle and other vehicles is crucial for road safety. Path prediction can assist the driver in having an enhanced perception of the road environment and of the intention of other neighboring drivers. In this paper, an advanced cooperative path prediction algorithm is presented. This algorithm gathers position, velocity, acceleration, heading, and yaw rate measurements from all connected vehicles to calculate their future paths. In addition, map data with regard to the road geometry and, in particularly, the road curvature are used to enhance the path prediction algorithm. Comparative results of the path prediction, with and without wireless communications, are discussed. In addition, the algorithm is adapted for use in the emergency-electronic-brake-lights application. The results of this adaptation are also presented. This paper is another contribution in highlighting the advantages and, at the same time, the challenges of using communications among road users.

Using digital maps to enhance lane keeping support systems

The development of a system that can be used for a safe, reliable, highly available onboard lane keeping support system is a critical research topic. One of the most important functions in driver assistant systems is the detection of unintentional lane departures. Current lane departure warning systems focus mainly in the detection of lane markings using vision sensors, such as CMOS cameras. In order to increase accuracy and robustness of such systems the utilization of digital maps is necessary. The goal of combining camera and map data is to extend the road geometry in further distances and eliminate false alarms based on unintentional maneuvers caused by the driver. The overall system efficiency is increased furthermore by using also vehicle dynamics and road geometry calculated using radar data.

Unscented Kalman filter design for curvilinear motion models suitable for automotive safety applications

Research in automotive safety leads to the conclusion that modern vehicle should utilize active and passive sensors for the recognition of the environment surrounding them. Thus, the development of tracking systems utilizing efficient state estimators is very important. In this case, problems such as moving platform carrying the sensor and maneuvering targets could introduce large errors in the state estimation and in some cases can lead to the divergence of the filter. In order to avoid sub-optimal performance, the unscented Kalman filter is chosen, while a new curvilinear model is applied which takes into account both the turn rate of the detected object and its tangential acceleration, leading to a more accurate modeling of its movement. The performance of the unscented filter using the proposed model in the case of automotive applications is proven to be superior compared to the performance of the extended and linear Kalman filter.

Detection of maneuvers using evidence theory

The early detection of maneuvers performed by the driver is one of the most important tasks of a driver assistance system for accurately perceiving the road situation and thus warning the driver in time or intervening in the vehicle control to avoid the danger. By recognizing relationships between entities of the road environment the system can react with more efficiency to the current situation. This paper focuses on the investigation of methods regarding the identification of the maneuvering type not only for the host vehicle but also for the detected objects which are moving in the surrounding environment using the theory of evidence.

Situation refinement for vehicle maneuver identification and driver's intention prediction

In safety automotive applications the system must me capable of early recognizing the maneuvers performed by the driver and the intention associated with them in order to take preventive measures or trigger warning alarms. This is done by the situation refinement level in the fusion system that processes the data provided by the on-board sensors. By recognizing relationships between entities of the road environment the system can react with more efficiency to the current situation. For example the intention of a lane change, the detection of an overtaking maneuver, the estimation of the lane in which the detected vehicle is located, help the system to decide which action must be taken in order to prevent an unwanted situation. This paper focuses on the investigation of methods regarding the identification of the maneuvering type and the intention associated.

Prediction of unintentional lane departure using evidence theory

The scope of this paper is the development of algorithms for driving support systems for safe lane changing maneuvers (lane change assist system) and safe maintenance of the vehicles path (lane/road departure warning system). The proposed algorithm can predict unintentional lane changes before they are performed by the driver using information from multiple sources. The prediction of unintentional lane changes comes from data produced by various sensors installed on the vehicle (inertial sensors, radar and camera). The decision fusion algorithm is based on Dempster-Shafer theory. The paper analyzes vehicle kinematics, extracts distributed decision components from sensor data and investigates several set of information sources and their mass functions to be utilized in the decision fusion system. Finally, uncertainties of each source are modeled and included in the Dempster-Shafers theory as weights calculated adaptively and in real-time using heuristics and a-priori knowledge. The algorithms are validated in real world scenarios.

Multiple hypothesis tracking for data association in vehicular networks

The introduction of Vehicle to Vehicle (V2V) and Vehicle to Infrastructure (V2I) communications in Intelligent Transportation Systems of the future brings new opportunities and new challenges into the automotive scene. Vehicular communications broaden the information spectrum that is available to each vehicle, allowing the enhancement of existing applications and the introduction of new ones. Undoubtedly, the impact of this new technology in transportation safety, efficiency and infotainment is expected to be very important. A significant part of research in vehicular networks (VANETs) is dedicated to networking issues like routing and safety. However, perception systems which until now were based on onboard sensors only, need to incorporate the wirelessly received information in order to extend the situation awareness of the vehicle and the driver. This paper presents an algorithm for associating targets tracked from an onboard radar sensor with the position and motion data received from the VANET. The core of the algorithm is a track oriented multiple hypothesis tracker that is modified for incorporating information included in VANET messages. The algorithm is tested in real scenarios using two experimental vehicles and then compared with two other algorithmic approaches. One is using a simpler single hypothesis algorithm for association of VANET messages and the second is using only the onboard sensors for environment perception. As a result, the advantages of the Multiple Hypothesis Algorithm regarding association performance and the added value of wireless information in the perception system are highlighted.

Multi-Sensor Coordination And Fusion For Automotive Safety Applications

This paper focuses on the solution of the problem of (onboard moving vehicles) multiple sensor data fusion systems. The proposed application uses distributed architectures that operate with sensors or sensor systems and give redundant or complementary information for moving objects. This architecture ensures a modular approach allowing exchangeability and benchmarking using the output of individual trackers, whereas the fusion algorithm gives a solution to the track management problem and the coverage of wide perception areas. The test case is a multi-sensor configuration, which monitors the rear and lateral areas of traffic. Results from simulations and real data show that the given approach allows maintenance of the ID of objects and recognition of the vehicle environment with acceptable rates of false alarm and misses