GPS World - May 2008 - (Page 34)

AVIONICS & TRANSPORTATION | Vehicle Integration Å FIGURE 7 PAS Horizontal Position, Velocity Error & DOP Å FIGURE 9 Collision prediction, ped speed = 0.1 meters/second Å FIGURE 10 Collision prediction, ped speed = 0.4 meters/second Å FIGURE 8 Kalman Filter Residuals To get more accuracy, we recommend implementing DGPS with carrier-cycle recovery to achieve accuracies to at least a decimeter. This method' s performance would be similar to the precision exhibited by the kinematic survey system that provided the truth measurements for this test. This method also would be more resistant to multipath errors. There are tradeoffs with a required longer data message that will affect the robustness of the radio link, but based on our recent experience with the PAS radio we think it could be successfully implemented. Å COLLISION prediction test setup The position error from the carrier-smoothed CA-code Kalman filter averages 1.55 meters with the average HDOP of 2.78. Knowing the separation of the vehicle and multiple pedestrians to an average of 1.5 meters over the PAS' effective range of zero meters to about a half mile in all weather and behind visual screens is not achievable with any other position sensing approach. The 1.5-meter result indicates that the PAS system has a base error of 0.56 meter when scaled by the average HDOP of 2.78. Improvements in the algorithm performance are possible with more analysis and testing. Addition of the new civil frequency since these tests in 2005 and the addition of the Galileo system will push average HDOP down close to 1.0 in the next decade. Unsurprisingly, the greatest error source appears to be multipath. 34 GPS World | May 2008 Collision Prediction Tests With the program end rapidly approaching, we sought to test the collision prediction algorithms with our remaining resources. However, the tests possible at this point did not come close in rigor and extent to what we had planned to perform. To simulate a ped dart-out crash Ð the most challenging ped crash scenario Ð without endangering people, we suspended a cable between two poles of differing heights. A weighted P-unit attached to a pulley was released from the highest point of the cable and the time it took to cross a 13-foot lane was determined. When the pulley rolled correctly, the transit time was a consistent 2.5 seconds. The starting point from which a car accelerating to 10 mph would reach the cable when the P-unit was half way across the cable was determined. We also conducted trials with www.gpsworld.com http://www.gpsworld.com

Table of Contents Feed for the Digital Edition of GPS World - May 2008

GPS World - May 2008 Contents Out in Front Letters to the Editor Expert Advice GPS III Contract Award Now a Reality? Near-Space Location Boost RTK Crops Up in Precision Ag Safety Afoot 50+ Leaders to Watch Making a Difference with GPS Product Showcase Advertisers Index & Company Directory Marketplace Classifieds Seen & Heard

GPS World - May 2008

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.