GISdevelopment.net --> Application --> Utility
Methods for Improving the Accuracy and Reliability of Vehicle-borne GPS Intelligence Navigation
Fenglin Guo1, Yuesheng Ji2 and Guorong Hu3
1Project Officer, School of Civil & Environmental Engineering
College of Engineering, Nanyang Technological University, Singapore
Block N1-B2a-06, 50 Nanyang Avenue, Singapore 639798
2Lecturer, School of Civil Engineering
Northern Jiaotong University, Beijing, China
3Research fellow, GPS Centre, College of Engineering
Nanyang Technological University, Singapore
With the development of GPS-based navigation, computer and communication, intelligent vehicle navigation system will play a more important role in future transportation systems. Vehicle-borne GPS can provide positioning and navigation information quickly and accurately at relatively low cost. GPS has made a significant impact on almost all positioning and navigation applications including vehicle-borne GPS intelligence navigation system (VINS) (Allison, et al., 1999; Haddad, et al., 1999; He et al., 1998; Hu et al., 2000). However, GPS alone is insufficient to maintain continuous positioning because of inevitable obstructions caused by buildings and other natural features. When GPS signals are blocked or lost, the precision of positioning will be reduced to unacceptable level. Therefore, it is necessary to improve the accuracy and reliability of VINS.
This paper presents three methods to improve the accuracy and reliability of VINS, viz., dead reckoning (DR), differential GPS (DGPS) and map matching. The advantages and disadvantages of these augmentation methods are analyzed. The results of a real application and simulation experiments are also provided. Further suggestions to assure the accuracy and reliability of VINS are also proposed.
Dead Reckoning (DR) Method
Dead reckoning method determine a vehicle’s position relative to an initial location by integrating measured distance increments and directions of travel (Hong, 1997). The distance increments are measured using a distance sensor. The directions can be derived through a course sensor. When there GPS signals are degraded, the position of a vehicle at ti epoch can be determined from the direction angle (α ) and distance (D) components:
(X0, Y0) the initial position at t0
(Xi, Yi) the position at ti
Dk, αk distance and direction from (X0, Y0) at t0 to (Xk, Yk) at tk epoch respectively.
The course sensor could be derived from geomagnetism, gyroscope or using information from the difference between the velocity of the left tyre and right tyre. The distance sensor may be tapped from the vehicle’s odometer or from a velocity sensor. The navigation accuracy of DR is a function of the distance traveled. Longer distances tend to incur greater accumulated errors. Errors of DR are mainly caused by characteristics of the sensors and from environmental factors such as terrain and uneven tyre pressures. Hence, DR per sec cannot be used over a long period. Navigation system which combines measurements from both DR and GPS system can mitigate the errors by continuously calibrating DR sensors by acquired GPS positions
Another mode of DR system is INS (Inertial Navigation System) which can continuously provide direction and acceleration (Yuan, et al., 1993). Starting from a known position, INS uses the variations in positions to determine the trail of a vehicle. Errors of INS increase with the square of time. Hence, INS alone has its limitations. However, a combined GPS and INS solution could overcome shortcomings of each other and is an effective method for providing continuous and precise navigation for vehicles. Examples can be seen from literature of Allison, et al., (1999).
Differential GPS (DGPS) Method
VINS belongs to a class of real time kinematic positioning whose precision is relatively low. In order to increase its precision, one method is to introduce differential GPS (DGPS) technique. DGPS can reduce or cancel error sources such as satellite clock bias, atmosphere delays, orbit bias. According to the different modes of operation, DGPS can be divided into three classes: position-based DGPS, pseudorange DGPS and carrier phase DGPS (Wang, et al., 1996; Hofmann-Wellenhof, et al., 1997). The principles are basically the same but corrected sophistication and precision levels of each technique are quite different. In VINS, position-based DGPS and pseudorange DGPS are usually used.
Fig. 1 shows that the precision of VINS can be improved to a meter level by using position or pseudorange DGPS (Wang, et al, 1996). Therefore, DGPS is a relevant method to increase the precision of VINS. With improved positions, the calibration of the sensors of DR system will have consequential improvements as well. The effect of using DGPS method to improve the performance of GPS within the urban environment is documented in Allison, et al., (1999).
Map Matching Method
The positions of a vehicle determined by GPS/DR or DGPS/DR could be displayed in electronic map. Because errors exist in both positions acquired by GPS/DR and also in digital maps, it is not possible to ensure that the positions of the vehicle register properly on a digital map. The result of this is that a vehicle may be seen to be moving over a building or into the sea. To avoid this phenomenon, map- matching method can be used to improve the displayed precision of vehicle over an electronic map. The principle of map matching method is to ensure that a position is snapped or matched to the nearest street. However, a street network can be quite complicated especially when there are several crossroads.
Determination of the correct street is not entirely straightforward. One such algorithm is proposed by Yi et al., (1998).
Figure 2 presents a section of a network and one point (P(X,Y)) representing a vehicle’s position. The position does not register correctly to a street. To find which street it belongs to, a circle with search radius R is drawn. With experience, a suitable value for R will be used. In this area, the objective is to find all streets that satisfy the following condition:
Distance (D) between street and P point is shorter than R.
There exists two possible cases:
According to above principle, a study was done with VINS for Wuhan City in China. Figure 3 shows the effect without using map-matching method. Some vehicle trajectories are outside the streets. Figure 4 demonstrates the visual improvement of the map matching method.
Conclusions and Recommendations
Three methods for improving the accuracy and reliability of VINS have been described. These methods are not independent each other. Unfortunately the additional augmentation will increase the cost. Moreover, in the course of developing VINS, several issues pertaining to the accuracy and reliability of VINS were also noted:
|© GISdevelopment.net. All rights reserved.|