Use Of Vision Sensors And Lane Maps To Aid Gps/Ins Under A Limited Gps Satellite Constellation

One major issue in implementation of a GPS/INS navigation system is the decrease in positioning performance in urban canyons. An urban canyon is any location where GPS satellite signals are blocked or corrupted by tall buildings. Tall buildings surrounding a GPS antenna will cause a masking affect on the antenna. Satellites that are at a low elevation angle will be blocked by buildings. Many of the signals that actually arrive at the antenna are corrupted by delays caused by multipath. A closely coupled GPS/INS navigation system with fault detection and exclusion can be used to combat the issue of multipath in urban areas. This paper will present a method of using vision measurements and a lane map to constrain the navigation system and thus improve observability.Many modern GPS/INS navigation systems use a closely coupled architecture. A closely coupled navigation systems refers to a navigation systems that uses the pseudorange and pseudorange-rate measurements provided by a GPS receiver. A loosely coupled navigation system refers to a system that uses only the position and velocity reported by the GPS receiver. One advantage of a closely coupled architecture is its ability to provide limited IMU corrections while receiving measurements from less than four satellites. A loosely coupled architecture will provide no IMU corrections if the GPS receiver fails to track four or more satellites. Another advantage of the closely coupled system is the ability for the system designer to incorporate intelligent measurement rejection to reject bad pseudoranges and pseudorange-rates.A closely coupled GPS/INS system with fault detection and measurement rejection can be used in an urban environment to mitigate navigation errors. The fault detection and measurement rejection will insure that bad pseudoranges and pseudorange-rates will not be used to compute the navigation solution. Furthermore, a complex elevation mask can be used to reject measurements from satellites that are believed to be currently blocked by buildings.One issue with pseudorange measurement rejection is the possibility of loss of observability. If the number satellite measurements used falls under four, then the GPS/INS system will not be fully observable. Also, when using a limited number of satellite observations, the observability of the GPS/INS system is heavily affected by the geometry of the satellites used.This paper proposes a method to increase observability of a GPS/INS system operating under limited satellite coverage. Extra range measurements from vision sensors are used to supplement the GPS's pseudorange and pseudorange-rate measurements. Both LiDAR and camera measurements are used to measure a vehicle's lateral position in its current lane. The vision measurements provide local based positioning based of the lane. A map of the lane is used to relate the vision's local positioning and the GPS's global positioning. Also, constraining the navigation system's height above the lane map is used to further aid observability.In order to test the performance of the navigation filter, real data from the NCAT test track in Opelika, Alabama will be used. The NCAT track has plenty of open sky; therefore, the solution using a full constellation of GPS satellites can be compared to the solution using only two GPS observations.