An Information Fusion Method for Vehicle Positioning System Yi Yan, Che-Cheng Chang and Wun-Sheng Yao Abstract Vehicle positioning techniques have a broad application in advanced driver assistant system (ADAS). The research of positioning information fusion methods is highly requested. In this paper, we propose a novel concept of information fusion method for positioning system, which aims to enhance not only the accuracy but also the reliability of positioning system. The system integrates several different sensing technologies, including GPS, inertial measurement unit (IMU) and odometer to calculate the position. It also uses the position of each environmental characteristic to reduce the error. The environmental characteristics are provided by lane departure detection system, traffic single detection system and road edge detection system. Finally, experimental results demonstrate the effectiveness of this positioning system. I. INTRODUCTION In urban areas, the reflected signals and the obscuration of signals make GPS derived position fixes less accurate. In order to combat these problems, the inertial navigation system (INS) is designed to complement the reliability and accuracy of the GPS, the integration of GPS and IMU provides a complete position solution with higher data rate. However, it will lead to lots of residual drifts if inertial sensor biases are not well estimated. In this paper, we propose a novel concept of positioning system, which aims to enhance not only the accuracy but also the reliability of positioning information. More specifically, our system is implemented via different sensing technologies, that is, GPS, IMU, odometer and environment feature detection and recognition. By incorporating these sensing technologies, the positioning system can fuse distinct sensing information to make the optimal positioning information according to different scenarios. Thus, no matter where the is, our positioning system can steadily provide the optimal positioning information. II. SYSTEM ARCHITECTURE Fig. 1 shows the concept of our positioning system. First, the raw data from GPS receiver is as the initial input. Then, we fuse relative positioning information provided by odometer and IMU and absolute positioning information provided by GPS and LiDAR to make the optimal positioning information. More specifically, the absolute data is as the measure part, and the relative data is as the predictive one. For example, by comparing heading angle information of GPS with IMU, information from IMU can be used to compensate if the GPS information is invalid, the fusion of GPS and IMU can cover the shortage of low data rate with GPS, the cumulate errors from IMU also can be eliminate. Fig. 2 shows the prediction and filtering concept. On the other hand, the most important design of this paper is that we also apply environment feature detection and recognition to obtain absolute positioning information. Fig. 3 shows that we use LiDAR to estimate the distance and angle relative to the and traffic sign. By searching HD map data, we can get the current absolute position. Thus the accuracy of position can be enhance while the system can detect some feature which is already store in database. Since we mainly focus on how to fuse positioning information, we will not illustrate the image processing skill detailedly. Figure 1. The architecture of our positioning system Figure 2. Vehicle works by filtering the raw data and calculate the displacement and direction of
Figure 3. LiDAR works by detecting environment feature and recognition In ideal state, the LiDAR with HD map will always provide accurate position of the. However, in real state, there are still some errors in detecting procedure and HD map. In order to correct the above situation, our method shown in fig. 4 also fuses the dynamic with current parameters and the estimated position from LiDAR to obtain the optimal solution. More specifically, we use weighting concept to determine to trust which information in distinct scenario. Thus, positioning error can be significantly reduced. Here, we start to illustrate our algorithm with figure 4. First, we define some variables for our algorithm (lines 1-10), e.g., ap gps is the absolute positioning information provided by GPS receiver, ap LiDAR is the absolute positioning information provided by LiDAR unit, rp fusion is the relative positioning information provided by fusion LiDAR unit and, ap est is the absolute positioning information estimate by our system. Upon receiving the absolute positioning information provided by GPS, the system calculates heading angle and updates measurement element of (lines 11-12). Upon receiving relative angle rate information provided by IMU, the system estimates heading angle and updates prediction element of (lines 13-17). On the other hand, upon receiving the relative displacement information provided by odometer, the system estimates ap est (lines 18-19). Upon detecting an obstacle, the system searches database for this feature and estimates rp fusion. Then, according to ap LiDAR and rp vm, the ap est will be updated (lines 20-24). Figure 4. Information fusion procedure
III. EXPERIMENT In this section, we illustrate how to realize our positioning system. We use GARMIN 18x-5Hz GPS, MicroStrain s 3DM-GX5-45 IMU, low cost odometer mounted in golf car and Velodyne s VLP-16 3D LiDAR to create our prototype. A RTK-GPS is also mounted on golf car, which was used as ground truth to assess the accuracy of positioning system. Fig. 5 shows the prototype of the positioning system. We drive the golf car through the L-turn and through lane and then compare information from positioning system proposed in this paper with the information provided by RTK- GPS receiver. Figure 5. The prototype of our positioning system Figure 6. Result of the L-Turn experiment without considering in this paper Figure7. Result of the L-Turn experiment with considering in this paper Table I. Results of the experiment in L-Turn Result of the experiment in L-Turn Using Unused Position Error(average) 0.14m 0.24m Position Error(maximum) 0.68m 0.96m
Fig. 6 and fig. 7 show the L-Turn trajectory through the administrative area of Automotive Research & Testing Center (ARTC). The blue trajectory is ground truth provided by RTK- GPS receiver, and the green trajectory is provided by our system. We can see that the result of the experiment using in this paper provides more accurate positioning information (Table I). Fig. 8 and fig. 9 show the through lane trajectory through the administrative area of Automotive Research & Testing Center (ARTC). The blue trajectory is ground truth provided by RTK-GPS receiver, and the green trajectory is provided by our system. Obviously, the green trajectory in figure 9 is more stabilized and accurate. The detailed experiment data is shown in Table II. Figure 8. Result of the through lane experiment without considering in this paper Figure 9. Result of the through lane experiment with considering in this paper Table II. Results of the experiment in through lane Result of the experiment in through lane Using Unused Position Error(average) 0.17m 0.19m Position Error(maximum) 0.42m 0.59m
IV. CONCLUSION In this paper, we propose a novel concept of positioning system by fusing the information from the, LiDAR and HD map. Thus, high accuracy positioning can be maintained from this information fusion method during GPS outages. The concept with enhances not only the accuracy but also the reliability of positioning information. According to the experiment results, we can observe that our positioning system works. In the future, we intend to design a new algorithm for our architecture to estimate more accurate positioning information. REFERENCE [1] P. T. Shaw and W. Pettus, An integrated approach to electronic navigation, OCEANS 2000 MTS/IEEE conference and exhibition, RI, USA, pp. 299-308, 11 Sep. 2000. [2] P. K. Tseng, M. H. Hung, P. K. Yu, S. W. Chang, and T. W. Wang, Implementation of an autonomous parking system in a parking lot, 2014 world congress on intelligent transport systems, Detroit, Michigan Sept. 2014. [3] G. Dissanayake, S. Sukkarieh, E. Nebot, and H. F. Durrant-Whyte, The aiding of a low-cost strapdown inertial measurement unit using constraints for land applications, IEEE transactions on robotics and automation, vol. 17, iss. 5, pp. 731 747, 2001. [4] C. C. Chang, P. K. Tseng, W. S. Yao, and C. Y. Lee, Cooperative Vehicular Surrounding Sensing System, 23rd ITS World Congress, Melbourne, Australia, Oct. 2016. [5] H. A. Karimi, J. G. Benner, and M. Anwar, A for navigation experience sharing through social navigation networks (SoNavNets), 2011 IEEE international conference on information reuse and integration (IRI), NV, USA pp. 557-560, Aug. 2011. [6] B. F. Wu, C. C. Kao, Y. F. Li, and M. Y. Tsai, A real-time embedded blind spot safety assistance system, International journal of vehicular technology, vol. 2012, pp. 1-15, 2012. [7] K. Lee and H. Peng, Evaluation of automotive forward collision warning and collision avoidance algorithms, Vehicle system dynamics, vol. 43, no. 10, pp. 735-751, 2005. [8] K. Kozak, J. Pohl, W. Birk, J. Greenberg, B. Artz, M. Blommer, L. Cathey and R. Curry, Evaluation of lane departure warnings for drowsy drivers, Proceedings of the human factors and ergonomics society 50th annual meeting, San Francisco, California,, pp. 2400-2404, 2006. [9] H.-J. Chu, G.-J. Tsai, K.-W. Chiang and T.-T. Duong, GPS/MEMS INS data fusion and map matching in urban areas, Sensors, vol. 13, iss. 9, Aug. 2013. [10] J. Wendel, O. Meister, C. Schlaile and G. F. Trommer, An integrated GPS/MEMS-IMU navigation system for an autonomous helicopter, Aerospace science and technology, vol. 10, iss. 6, pp. 527 553, 2006. [11] D.M. Bevly, Global positioning system (GPS): A low-cost velocity sensor for correcting inertial sensor errors on ground s, Journal of dynamic systems, measurement, and control, vol. 126, iss. 2, pp. 255 264, 2004. [12] S. Sukkarieh, E.M. Nebot, and H. F. Durrant-Whyte, A high integrity IMU GPS navigation loop for autonomous land applications, IEEE transactions on robotics and automation, vol. 15, iss. 3, pp. 572 578, 1999. [13] S. Godha and A. M. E. Cannon, GPS/MEMS INS integrated system for navigation in urban areas, GPS solutions, vol. 11, iss. 3, pp. 193-203, 2007. [14] P. T. Shaw and W. Pettus, An integrated approach to electronic navigation, OCEANS 2000 MTS/IEEE conference and exhibition, RI, USA, pp. 299-308, 11 Sep. 2000. [15] Global Positioning System Wikipedia, https://en.wikipedia.org/wiki/global_positioning_system. [16] Inertial measurement unit - Wikipedia, https://en.wikipedia.org/wiki/inertial_measurement_unit [17] Odometer - Wikipedia, https://en.wikipedia. org/wiki/odometer. [18] A Tutorial and Elementary Trajectory Model for The Differential Steering System, http://rossum.sourceforge.net/papers/diffsteer/diffsteer.html.