NAVIGATING SYSTEM VEHICULAR
BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to vehicle navigating systems and more particularly, to vehicular navigation systems that use physical and distinctive virtual badges in order to improve the status information obtained from the navigation system. 2. Description of the Prior Art In general, there has been a growing interest in navigation systems and their applications. More specifically, navigation systems provide vehicle status information that is useful for simultaneously monitoring numerous vehicles as well as controlling the route and distribution of vehicles. Various navigation systems use a variety of instruments to monitor the movement and position of vehicles. Traditionally, the position and speed of a transit vehicle was determined through the use of rotation sensors mounted on the wheel, which counted the number of revolutions of the wheel and fractions thereof. An odometer would multiply the number of rotations by the diameter or circumference of the wheel to obtain a distance calculation. In addition, a tachometer could be used to multiply the counting of wheel rotations over a specific time interval by the diameter or circumference of the wheel and divide the result between the time interval to obtain the average speed of the vehicle over the time interval . However, these instruments depend on the mechanical contact between the vehicle and the operating environment and are therefore subject to loss of transmission power and errors. Some modern navigational systems, such as Global Positioning Systems, solve some inherent problems with traditional instruments, but they also have disadvantages. For example, the operational capabilities of Global Positioning Systems are limited by line-of-sight coverage, obstacles, foliage and terrain characteristics. Other navigational systems follow an almost-autonomous approach to obtain the measurement of the speed and position of the vehicle. For example, U.S. Patent No. 5,332,180, which is incorporated herein by reference, discloses a vehicular navigation system that utilizes inertial on-board sensors to provide vehicle speed and position information. Alternatively, an estimation system could be used to provide dynamic vehicular information. The autonomous nature, without contact to perceive the esteem or inertial form provides an attractive method to measure the state characteristics of the vehicles. However, these systems accumulate errors over time that must collapse to provide accurate, reliable and useful information. Various map comparison techniques provide mechanisms to reduce errors in navigation systems. Specifically, the vehicle state information of navigational instruments can be compared with known data to provide estimates of location and speed in order to reduce errors within the system. U.S. Patent No. 5,332,180 discloses the use of physical signs and / or active guides to carry out the required error collapse functions. The ability to reduce errors is related to the number of physical signs used in the system and errors can be reduced at a rate proportional to the number of physical signs present in the navigation system.
However, the use of numerous physical labels or active guides to reduce errors within the system increases system maintenance and hardware failure opportunities. In addition, traditional map comparison methods have inherent accuracy and limitations of execution. SUMMARY OF THE INVENTION The invention provides a vehicular navigating system where the measurement of the characteristics of the vehicle state is improved without the use of additional physical devices along the vehicular route. The vehicle navigational system according to the present invention preferably includes an inertial measurement unit to generate kinematic variables from the vehicular movement and a navigator to calculate the measured state characteristics of the vehicle from the kinematic variables. The vehicle navigational system additionally includes a converter for generating a descriptive identification of the kinematic variables and a recognition pattern analyzer such as a neural network for comparing the descriptive identification with a reference identification derived from a recognition of the profile of the vehicular route. . The comparison of identifications allows the system to use virtual badges to more accurately estimate the vehicle's state characteristics. The vehicular navigation system may additionally include a plurality of physical indicia located adjacent to the vehicular route and a distinctive reader on board the vehicle to provide instantaneous status characteristics and collapse the accumulated errors within the system. The vehicular navigating system may include a receiver of the Global Positioning System to provide absolute state characteristics. The vehicular navigation system can additionally include a filter to more accurately estimate errors within the system and allow a more accurate count of accumulated errors. The proportion of a neural network to generate virtual badges improves the accuracy of the vehicular navigation system without increasing the number of physical badges in the system. It is preferred to reduce the number of physical marks in order to reduce the maintenance of the system and the possibility of failure of one or more of the physical badges. BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a plan view of a main vehicular route that has a lateral route;
Figure 2 is a diaphram of functional block of the vehicular navigational system according to the invention; Figure 3a is a schematic diagram of a neural network that can be used to estimate the state characteristics of the vehicle; Figure 3b is a schematic diagram similar to Figure 3a of a recurrent neural network; and Figure 4 is a flowchart of a process to collapse errors within the vehicular navigation system through the use of virtual badges. DESCRIPTION OF THE PREFERRED MODALITIES As shown in Figure 1, the vehicle navigational system 10 according to the invention uses a plurality of physical badges 14 to provide positive position information of a vehicle 12 in a plurality of locations on a vehicular route. This information can be used by the vehicular navigation system 10 to collapse the accumulated errors present within the system. The vehicle navigational system 10 according to the invention additionally uses a plurality of virtual badges 16 to provide a plurality of additional points where accumulated errors can collapse. The virtual badges 16 are calculated by a recognition pattern analyzer such as a neural network 22 and can be used to more accurately estimate the state characteristics of the vehicle 12 in a plurality of positions on the vehicular route 13. The movement of the vehicle 12 on the vehicular route 13 is monitored by an inertial measurement unit 18 on board the vehicle 12. The inertial measurement unit 18 preferably includes at least one gyroscope and at least one accelerometer. The inertial measuring unit 18 can obtain kinematic variables which include an instantaneous acceleration measurement and a measurement of the instantaneous rotational speed of the vehicle 12. As shown in Fig. 2, the output of the inertial measurement unit 18 is sent to a navigator 20. Navigator 20 can integrate the output of the accelerometer to calculate status characteristics such as vehicle speed and position measurements 12. Navigator 20 can provide additional status features, including altitude information, by integrating the rotational speed of the vehicle 12. The movement of the vehicle 12 along the vehicular route 13 can be subsequently approximated by the inertial measurement unit 18 and the navigator 20. However, the errors in the data provided by the inertial measurement unit 18 and the navigator 20 will increase as a function of time. The vehicle navigational system 10 according to the invention can use a variety of methods to reduce the errors it accumulates within the system. Specifically, the vehicular navigation system 10 may use a Global Positioning System or a Differential Global Positioning System to update the position of the vehicle 12. As shown in Figure 2, the vehicular navigation system 10 may include a Positioning System receiver. Global 15 for receiving absolute state characteristics from a Global Positioning System 11. The vehicular navigation system 10 may additionally include a plurality of physical indicia 14 located adjacent to the vehicular route 13 in a variety of locations thereof. . The system employs the physical indicia 14 to provide instantaneous, accurate status information along the vehicular route 13. As shown in Figure 1, the physical indicia 14 are preferably positioned adjacent the switches 17 to provide reliable information of the vehicular route. Physical indicia 14 may include transponders that transmit information such as a distinctive number, identification of vehicular route 13, distance along the vehicular route
13 or latitude and longitude orientation. According to the foregoing, the vehicle 12 may also include a badge reader 30 such as a read / write transponder device for detecting the presence of physical badges 14 along the vehicular route 13 and obtaining instant information of the vehicular route. 13 from it. The information read by the badge reader
from the physical indicia 14 can be used to reduce the errors that have accumulated within the vehicular navigation system 10 by updating the vehicular state characteristics such as position within the vehicular navigation system 10 in a high degree of certainty. In particular, each physical tag 14 provides a greater error collapse structure within the vehicular navigational system 10. Accordingly, it would appear to be favorable to provide numerous physical tags 14 along the travel path of the vehicle 12 in a plurality of positions. to improve the accuracy of the navigation system. However, the use of physical hallmarks
14 increases the maintenance and installation time of the system as well as the possibility of hardware failure.
The vehicle navigational system 10 according to the invention uses a plurality of virtual badges 16 to provide additional points along the vehicular route 13 where errors accumulated within the system can be reduced. The virtual badges 16 are established in the system by means of a neural network 22. The neural network 22 determines the location of a plurality of virtual badges 16 along the vehicular route 13 through a recognition pattern. Initially, the neural network 22 can be trained to recognize the identifications corresponding to a plurality of positions on the vehicular route 13 transited by the vehicle 12. A recognition profile of the vehicular route 13 can be used to train the neural network 22. The The recognition profile of the vehicular route 13 preferably includes a plurality of kinematic variables corresponding to a plurality of positions on the vehicular route 13. The kinematic variables of the recognition profile of the vehicular route 13 are introduced in a first converter 21 where the variables kinematics are grouped into a plurality of reference identifications. The neural network 22 is trained to associate each reference identiication with a specific position on the vehicular route 13. The reference identifications are preferably associated with curves and switches 17 where the output of the inertial measurement unit 18 is strong and In addition, each reference identification can include a sequence of kinematic variables, thereby allowing the neural network 22 to differentiate between similar identifications on the vehicular route 13. The vehicular navigation system 10 calculates additionally a plurality of descriptive identifications as the vehicle 12 travels along the vehicular route 13. In particular, the kinematic variables of the inertial measurement unit 18 are applied to the first converter 21 where a plurality of identifications is generated. descriptors. The descriptive identifications are applied to the neural network 22 to compare each descriptive identification with the plurality of reference identifications. The neural network 22 attempts to compare each descriptive identification obtained from the output of the inertial measurement unit 18 with a reference identification obtained from the recognition profile of the vehicular route 13. The second converter 23 can determine the location of the virtual badge 16 and estimate by this the state characteristics of the vehicle 12 once the neural network 22 compares the descriptive identification with a reference identification. The use of virtual badges 16 provides a lower error collapse structure within the vehicular navigation system 10 without the use of additional physical badges 14. The output of the neural network 22 is applied to a second converter 23. The second converter 23 is configured to calculate the estimated state characteristics corresponding to the identified virtual badge 16. The first converter 21, the neural network 22 and the second converter 23 can be implemented in a processor on board the vehicle 12. Figure 1 shows an example of the use of the vehicular navigation system 10 in a railway environment. A switch 17 is located along the vehicular route 13 and a vehicle 12 passing over the commutator can experience a sudden lateral movement. Preferably, the sudden lateral movement will be included in the recognition profile of the vehicular route 13. Accordingly, the first converter 21 will generate a unique reference identification corresponding to the location of the switch 17 on the vehicular route 13 during the training of the neural network 22. In addition, the lateral movement of a vehicle 12 passing over the switch 17 can be detected by the inertial measuring unit 18 on board the vehicle 12. The first converter 21 generates an individual descriptive identification corresponding to the lateral movement of the vehicle 12 and the neural network 22 compares the descriptive identification with the reference identifications developed from the recognition profile of the vehicular route 13. The neural network 22 compares the descriptive identification with the corresponding reference identification and identifies a virtual badge 16 corresponding to the location of the utador 17 on the vehicular route 13. The output of the neural network 22 corresponding to the identified virtual badge 16 is subsequently advanced to the second converter 23. The second converter 23 then generates estimated status characteristics that include estimates of the vehicle's location 12 that can be used to collapse errors within the vehicular navigation system 10 without the use of additional physical marks 14.
The neural network 22 shown in Figure 3a can be used within the vehicular navigation system 10. The reference identifications corresponding to the recognition profile of the vehicular route 13 are applied to an input layer 24 while the neural network 22 is in a modality of training. In addition, the descriptive identifications determined from the movement of the vehicle 12 are applied to the input layer 24 when the neural network 22 is in a processing mode. All identifications are received by a first set of processing units 27a or nodes within the input layer 24. The output of each processing unit 27a is applied to a second set of processing units 27b, which constitute a hidden layer 25, through a first set of connection links 28a. The output of each processing unit 27b within the hidden layer 25 is applied to a third set of processing units 27c, which constitute an output layer 26, through a second set of connection links 28b. The neural network 22 carries out the recognition pattern through connection links 28a, 28b between the layers 24, 25, 26. The output layer 26 sends an output identification to a second converter 23 in response to the comparison of a descriptive identification and a reference identification and the identification of output corresponds to the identified virtual badge 16. Preferably, the neural network 22 can include a recurrent neural network in which the output of one of the layers is fed back to the entrance of the same or of a previous layer within the neural network 22 as shown in Figure 3b. The output is subsequently combined with the normal input of the layer or a previous layer to provide knowledge of the current and past entries. A recurrent neural network may be necessary to distinguish similar identifications within the recognition profile of the vehicular route 13 and correctly calculate the virtual flags 16. The recurrent neural network can estimate the location and speed of the vehicle 13 with an increased degree of certainty to the recognize the descriptive identifications within a sequence of identifications. In addition, the use of a recurrent neural network allows the vehicular navigation system 10 to estimate future identifications based on the current entry and the last entry. The neural network 22 shown in Figure 3a can be modified to form a recurrent neural network as shown in Figure 3b. Specifically, the neural network 22 operating with a recurrent neural network would include a plurality of feedback connection links 29 as shown in Figure 3b. The feedback connection links 29 allow the individual layers within the neural network 22 to process current and past data to improve the recognition accuracy of the neural network 22 and allow the neural network 22 to predict future identifications of the vehicle 12 on the vehicular route 13. The flow diagram of Figure 4 shows the process for using neural network 22 to calculate virtual badges 16 as minor error collapse structures. In the first step 40, the inertial measuring unit 18 measures the kinematic variables corresponding to the movement of the vehicle 12. The first converter 21 in the next step 42 calculates a descriptive identification from the kinematic variables. In step 44, the neural network 22 compares the descriptive identification with a plurality of reference identifications generated from the recognition profile of the vehicular route 13 and identifies a virtual badge 16 corresponding to the identifications compared. In step 46, the second converter 23 calculates estimated state characteristics from the information corresponding to the identified virtual badge 16. The state characteristics of the receiver of the Global Positioning System 15, the navigator 20, the second converter 23, and the flag reader 30, can subsequently be applied to a recursive estimation filter as shown in Figure 2 to generate the filtered state characteristics. The recursive estimation filter used by the vehicle navigational system 10 is preferably a Kalman filter. The Kalman filter 32 combines the state characteristics of the receiver of the Global Positioning System 15, the navigator 20, the second converter 23 and the badge reader 30 in order to estimate the errors within the system. The Kalman filter 32 can also eliminate the past observed data since the previous data and the current data of the Global Positioning System receiver 15, the navigator 20, the second converter 23 and the flag reader 30. are known. filter 32 Kalman improves the overall efficiency and accuracy of the vehicular navigation system 10. The Kalman filter 32 attempts to collapse or reduce errors within the vehicular navigation system 10 by processing the redundant state characteristics that are measured simultaneously from a plurality of sources. The Kalman filter 32 can be configured to assign independent weights to the state characteristics of each source depending on the level of confidence in the accuracy of the source. In general, the instantaneous state characteristics provided by the physical flag 14 and the flag reader 30 are considered as a strong constraint and define a larger error collapse structure. The information of the virtual badges 16 is considered to be a delta correction for the information of the physical badge 14 and can be defined as a minor structure correction. The filtered status characteristics of the Kalman filter 32 are applied to the navigator 20. The navigator 20 collapses the errors accumulated within the vehicular navigation system 10 based on the filtered status characteristics. An output of corrected errors from the navigator 20 can be immediately advanced to a central control equipment 34 for vehicular monitoring or traffic control, or a deployment on the vehicle 12 for on-board monitoring. The vehicle navigational system 10 continuously monitors the movement of the vehicle 12 on the vehicular route 13 through the inertial measurement unit 18 and reduces the errors with reference to the virtual badges 16 and the physical badges 14.
Although the specific embodiments of the invention have been described with d € italles, it will be appreciated by those skilled in the art that various modifications and alternatives to those details could be developed in light of the total teachings of the exhibition. According to the foregoing, the particular arrangements set out are intended to be illustrative only and not limiting to the scope of the invention, which is given by the full extent of the following claims and the equivalents thereof.