Debiao Lu

Methods for certification of GNSS-based safe vehicle localisation in driving assistance systems

Global Navigation Satellite Systems (GNSS) are now widely used in road transportation systems for vehicle navigation, fleet management, satellite-based road traffic monitoring and others, which cover 38% of all the GNSS applications. GNSS can also be used to leverage precise and safe locations in situations with potential harm to humans or damage to a system or environment, which is called safety-related applications. A user case can be demonstrated in advanced driving assistance systems (ADAS), GNSS receiver is applied as part of the satellite-based vehicle localisation unit (SatVeLU) to deliver safe locations continuously. It is highly necessary to certify this kind of localisation unit product to make sure SatVeLU can be used in safety-related applications. This paper analyses the quantifiable characteristics for GNSS-based safe vehicle localisation performance, an UML-based formal method as attribute hierarchy is used to describe the terminologies for it. With a clear interpretation from performance concept to quantifiable characteristics and a pre-calibrated mobile reference measurement system (MRMS), the performance characteristics can be validated through the proposed validation methodology. The validation methodology involving related international standardisation organisations, governmental institutes, agencies, accreditation bodies as well as laboratories together can deliver a complete certification process. This proposed certification process will potentially enable GNSS-based safe vehicle localisation units to be certified and finally applicable for safety vehicle localisation function in road transportation.

Particle Filter technique for position estimation in GNSS-based localisation systems

The usage of filter techniques for position estimation for safety-relevant purposes is an extensive field of research. Depending on the needed application for the Global Navigation Satellite System (GNSS) the state estimation can be achieved by several techniques. State estimation by means of Kalman Filter (KF), as well as Extended Kalman Filter (EKF) and Particle Filter (PF) have been developed and tested. Map-matching algorithms integrate the localisation data provided by GNSS with spatial road network data (also called “digital map”) to identify the correct line (or track) on which a vehicle is traveling and to determine the location of a vehicle within the line (or track). The goal of map-matching techniques is to exploit prior information contained in road or railway networks. However, incorporating digital map information within the conventional KF framework is not easy, because this constraint leads to highly non-Gaussian posterior densities that are difficult to represent accurately using conventional techniques. Since the PF approach presents no restrictions regarding non-linearity of models and noise distribution the velocity and heading measurement errors can be accurately modelled. The most significant advantages of the PF approach for map-matching application are: 1) PF approach provides a natural way for road map information to be incorporated into vehicle position estimation. 2) PF approach is capable of capturing multi-modal distributions. The selected PF-based location estimators presented here is oriented to work within an intelligent GNSS-based localisation system. The PF-based map matching techniques are presented in a mathematical ground and tests are performed, as part of a developed satellite localisation system based on artificial intelligence (AI) tools. In the railway domain the same integration can be performed by means of a “digital trap map” to identify the location within the tracks. A map-matching algorithm can be the key component of the data fusion to improve the performance of localisation systems that support the navigation function of intelligent transport systems (ITS).

Application of DBN for Assessment of Railway Intelligent Signal System Reliability

According to the variable structure characteristics of railway intelligent signal system (RISS) with different railway station scale, a new reliability assessment method based on Dynamic Bayesian Networks (DBN) is studied. A comparison between DBN model and probabilistic model is studied to verify the accuracy and correctness of DBN model. Based on DBN model, the static gates analyzing results deliver a calculation with no error, while the spare gate analyzing results deliver a calculation with a tolerable error that leads to more strictly and credible calculations. Meanwhile, this paper analyzes reliability indexes of RISS with four different railway station scale. The results show that: when the railway station scale increases, the reliability of RISS decreases, which has little impact on the ranking of the components’ Birnbaum importance factor and diagnostic importance factor.

Integrity of GNSS-based Train Positioning: From GNSS to sensor integration

Accurate and safe train position determination is of great importance for railway systems. Additionally, it has become one of the biggest challenges for the Safety-of-Life (SoL) services in railway transport systems using GNSS techniques. The critical performance requirements of these services promote the users to develop safe Train Positioning Systems (TPSs), in which integrity has to be highly concerned to real-timely monitor the quality of the TPS results. This paper presents a solution to the problem of integrity monitoring in GNSS-based train positioning with a global prospect. Different from the RAIM (Receiver Autonomous Integrity Monitoring) approach, a novel hierarchical architecture is proposed to expand the RAIM concept to the TPSAIM (Train Positioning System Autonomous Integrity Monitoring) domain, where integrity monitoring is closely correlated with both GNSS navigation computation at the local level and sensor integration at the global level. Integrity monitoring and fault detection with respect to the integration of GNSS and the Dead Reckoning (DR), which can be simply implemented with odometer and gyroscope, are investigated with different coupling structures. Results from simulations based on the field experiment illustrate characters of the proposed TPSAIM solution, and demonstrate its capabilities in terms of the effectiveness, coverage and flexibility with various TPS structures and operation conditions.

Accuracy analysis of BeiDou receivers for lane detection applications

The future of Intelligent Transportation Systems (ITS) relies on a properly constructed Control Transportation System (CTS) based on a dynamically accurate localisation system. In urban scenarios applications for lane detection, as part of intelligent highways, Global Navigation Satellite Systems (GNSS) receivers provide tangential and perpendicular locations for localisation systems, by means of trueness and precision in the dynamic frame of the lane. Results show that the circular error probable (CEP) approach is not a realistic representation of the dynamic behaviour of the receivers, so the new Mahalanobis Ellipses Filter (MEF) approach was tested and proven to be a better dynamic accuracy representation for ground vehicles. Also, the MEF results from collected data in urban scenarios in Beijings 3rd Ring Road present a significant representation of the potential applications for the several GNSS configuration tested, proving that a Multi-GNSS configuration can already properly work for lane detection applications. Conclusively, the results presented here show the MEF approach to be significantly better to test and validate lane detection applications. The separation into tangential and perpendicular deviations components allows a better technique to achieve the accuracy requirements description, for GNSS-based urban ground vehicle localisation system within a safety-relevant ITS.

Repeatability test method of GNSS for safe train localisation in real and simulated environments

Safe train localization is the key element to provide reliable Position, Velocity and Timing (PVT) information for Automatic Train Operation (ATO), and will enable future applications such as moving block, real-time safety margin estimation for adjacent trains, and many more. The safety-assured train localization under many repeated tests along the test railway line providing acceptable performance level will allow Global Navigation Satellite Systems (GNSS) as part of the localization unit to be adopted into the elements of ATO system to perform localization function. But under real railway operation environmental conditions, the accuracy performance shows strong variations; but using the simulated GNSS signals in the laboratory will have more consistent and repeatable test sequence for GNSS receiver testing. The GNSS receiver for the application of ATO is based on the performance requirements for the localization function in predefined environmental conditions, both simulation and real world test procedures including standard test definitions will enable the possibility to examine the repeatability of GNSS performance in acknowledged railway operation environment. Based on this, the test results will give the repeatability facts of the tested GNSS receiver, then the repeatability performance criteria can be set for later railway application in different environments. But for the performance evaluation of the repeatability property as part of the GNSS performance testing concept, accuracy can be seen as the foundation of all the other performance properties. Accuracy in both along-track and cross-track direction of the train should be evaluated under dynamic movement using bidirectional error models. The dynamic train localization accuracy is among one of the key tests for the GNSS receiver for safe applications in railway applications. This paper firstly describes the repeatability performance as part of the GNSS receiver quality for safe train localization in definition and also the relation among the major performance properties of a GNSS receiver under dynamic measurements. Secondly, the test scenario and the data collection strategy on an existent railway line is settled for both real world and simulated data analysis. Then the collected data is evaluated using the proposed dynamic accuracy evaluation method with the consideration of measurement uncertainty caused by the movement of the measurement object. Thirdly, the accuracy evaluation is based on the GNSS receiver as the system under test (SUT) and the mobile reference measurement system (MRMS) developed based on mature solutions and further developed platform in our research lab. The same datasets for real world accuracy evaluation are loaded into the Spirent GSS8000 for the repeatability performance evaluation. The test procedure is restricted to follow the standard test requirements in both railway and navigation societies. Finally, with the test results following the defined procedure and evaluation method in this paper, the repeatability performance criteria can be clearly shown for the level of safety application in railway train operation.

Safety Margin Estimation Using Risk Assessment Method for Satellite-Based Train Localization Unit

Safety as the key quality property among RAMS (reliability, availability, maintainability, and safety) demonstrates the most stringent performance in correspondence with the safety requirements and performance standards like EN 50126. Meanwhile, GNSS (Global Navigation Satellite Systems) are penetrating the railway now widely in non-safety related applications as passenger information, fleet management, etc. GNSS also have great potential for safety-related applications in railway such as the train location determination function, which the safety performance needs to be assured through hazard analysis and risk assessment process.The train location determination by satellite-based localization system is elevating the train control to the next level. The European Train Control System (ETCS) has being trying to implementing Level 3, the Chinese Train Control System (CTCS) has been implementing CTCS Level 3 low cost especially for secondary lines, and the U.S. is implementing train control systems under Positive Train Control (PTC) requirements. The train control system needs GNSS to provide more accurate location information of trains, more flexible and condensed trains on tracks with the consistency of still keeping the current safety level or even improve safety.Some researchers are trying to understand the performance of GNSS (GPS / EGNOS / Beidou) for railway applications from the fundamental accuracy level. A satellite-based train localization unit (SaLuT) as the entity to perform the train location determination function is to bring the GNSS accuracy evaluation up to safety integrity according to the safety requirements and standards for risk assessment. One of the key consequential result derived from the train location is the adequate safety margin. The safety margin, which can also be called as “safe braking distance”, is a margin indicated to rail traffic that would allow the train to stop with the application of normal service braking. The safety margin estimation quality and the risk of the safety margin shows the hazard rate for the safety margin estimation function performed by the designed localization unit SaLuT.This paper discusses the safety margin estimation method considering both GNSS accuracy and integrity assessment aspects of SaLuT, in accordance of the settled safety requirements of location determination function. To analyze the hazard of the safety margin estimation, a formal method is applied to model the SaLuT behavior and functions. The formal method based on stochastic Petri net enables the modeling process to include the GNSS receiver collected real data on the test track into it. The safety margin estimation method together with the risk assessment method using the real data can generate quantitative indicators to represent the localization function and safety margin estimation quality. The data used for the analysis is collected in the Qinghai-Tibet railway line from Golmud station to Ganlong station by SaLuT installed on a locomotive along the track. With the stochastic Petri net model and the systematic equation using the real collected data to estimate the safety margin based on the GNSS technologies, the SaLuT can be validated and verified for its hazard rates, which provides information for the safety cases in order to meet the industrial normative requirements.Copyright

GNSS-Based Train Trajectory Simulation System

Integrating Global Navigation Satellite System (GNSS) into railway application has a great potential because of its various advantages, such as lower cost, less trackside equipment, higher positioning accuracy, easier maintenance and so on. Railway system is a safety-critical system that requires high reliability, safety and real-time performance, so GNSS technology must be tested, verified and validated in railway system before putting into practical applications. However, due to the unavoidable restrictions and inconvenience of the railway field conditions, these tests cannot be accomplished on site. On this basis, this paper has developed a GNSS-based train trajectory simulation system which can provide GNSS data simulation of multi-train trajectory in multiple scenarios in order to support the tests and research of GNSS-based railway application, especially GNSS-based train localisation system and GNSS-based train control system.The GNSS-based train trajectory simulation system is based on the railway timetable (also called schedule), rolling stock information and digital track map. The paper firstly researches on the timetable that stores information of each train at each specified station, including arrival time, departure rime, track to be occupied, and connections to other trains. With the timetable simulation, the train’s trajectory can be generated using the information provided by the digital track map. The output trajectory data is mainly GGA sentence which is compliant with the National Marine Electronics Association (NMEA) 0183 standard. The paper also calculates the satellite visibility based on satellite ephemeris to simulate the number of visible satellites during the trajectory with changing time and space. All the information and data, such as timetable, speed/distance curve, distance/time curve, station track occupation state, can be visualized and updated in graphics and diagrams for better view. In addition, the train motion behavior of acceleration, cruising, coasting and braking can also be modelled in the system, as well as the driver’s behavior.The GNSS-based train trajectory simulation system has been realized using C# programming language in Microsoft Visual Studio 2010. And the field data of Shanxi coal railway transportation company railroad is used in the system. The simulation system is tested and the experimental results show that the developed simulation system can perform the expected functions, and provided data source for GNSS-based train localisation system. In addition, this simulation system has a good performance in compatibility and scalability.Copyright

Error bound estimation of cooperative vehicle localization using an integrity concept

The cooperative vehicle localization using DSRC (Dedicated Short Range Communication) is of great significance in many future transportation applications. An overlay function of DSRC has been explored with vehicular communication and it is with great potentials to achieve accurate and reliable vehicle cooperative localization with assistance from the GNSS (Global Navigation Satellite System) receivers. In order to establish a solid foundation for emergent implementations of the connected vehicles scheme, an effective error bound strategy for filtering-based estimation in cooperative localization using the GNSS and DSRC needs to be concentrated to bridge the connection between localization and extended functions. In this paper, a tight-coupled architecture for integrating GNSS and DSRC is investigated to realize cooperative localization, and a corresponding error bound estimation method is addressed. Based on the integrity concept, a confidential interval-based strategy is presented to constrain the uncertainties in filter estimation, and the combination of space alignment calibration and the additional safety margin further enhances the capability of the presented solution. The simulation results demonstrate the potential for safety-related applications under the connected vehicular environments.

Simulink based prototype for real-time intelligent GNSS-based localisation system

In the frame of the development of artificial intelligent (AI) based validation tool for Global Navigation Satellite Systems (GNSS) localisation systems MATLAB Simulink models for all the developed AI-based methodologies have been created. The combination of Artificial Neural Network (ANN) based validation tools with the Mahalanobis Ellipses Filter (MEF) methodology for accuracy-based data evaluation for quantitative and qualitative analysis of trueness and precision, as well as Particle Filter (PF) techniques for position estimation to aid the reference system result in significant improvements for GNSS-aided localisation system, allowing the construction of a GNSS-dependent reference system for when no independent reference is available. These additional elements are the bases for future intelligent GNSS-based localisation systems with both quantitative and qualitative validation tools. The proposed prototype constructed within System functions (S-Functions) for extension of MATLAB Simulink capabilities presents total interconnectivity in real-time, implemented using commodity microcomputers and providing an interface between the MATLAB Simulink implemented algorithms and the real world. In the presented prototype a Raspberry Pi Model B microcomputer running a modified Debian Linux distribution provides General Purpose and UART ports that enable the implementation of the low-level communication driver for the GNSS receiver. The prototype enables real-time application of the algorithms. The prototype includes a GNSS receiver, with a sampling position rate and constellation data at 1Hz. NMEA format provides the necessary data to be processed with the MATLAB Simulink model of the ANN-based validation tools and the PF-based estimator. All algorithms are executed in a Windows PC and the communication between the microcomputer and the MATLAB Simulink model is achieved via TCP Sockets, with interface S-Functions to handle the communication. These methodologies will enable future safety-relevant applications, such as on-board uncertainty evaluation; advanced driver assistance systems; and GNSS-based vehicle localisation with intelligent maps for track selective.