Continuous Navigation - Combining GPS with Sensor-Based Dead Reckoning - GPS World
Continuous Navigation
Combining GPS with Sensor-Based Dead Reckoning
GPS World

Pages | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8

Figure 6 Multipath effect. Reflected signals combine with the direct signal at the receivers antenna resulting in a position error.
Many algorithms simply switch between pure GPS and pure dead reckoning. With this approach, it is not possible to utilize the complementary advantages of GPS and dead reckoning where the weaknesses of one system are compensated by the strengths of the other system. Degraded positioning accuracy and unwanted position jumps are the result of this sub-optimal approach.

Figure 7 An enhanced Kalman filter optimizes the navigation solution according to the availability and quality of the GPS-derived position.
Weighted-Mix Solution. In contrast to the switching approach described above, a weighted mix of GPS and dead reckoning counters these disadvantages. The algorithmic approach, embedded into an enhanced Kalman filter (EKF), is graphically illustrated in Figure 7.

This approach eliminates multipath effects, position jumps, and distortions from jamming sources. Depending on the quality of the GPS signal (indicated, for example, by the number and distribution of visible satellites, dilution-of-precision value, etc.) and on the confidence level of the dead-reckoning signal (that is, how well the sensors are calibrated at the moment), a weighted mix of both GPS and dead reckoning is chosen to generate optimal results.

In practice, this approach can provide uninterrupted, reliable navigation results in the most-challenging urban environments including New York, Hong Kong, and Tokyo.

Automatic Calibration For accurate dead reckoning, calibration is required for the odometer pulses and the gyroscope. A well-designed GPS receiver with dead-reckoning capabilities applies a fully automatic calibration process during periods of very good GPS reception.

The process should be transparent, meaning that the car driver does not need to take any manual actions to perform a calibration such as setting parameter values or making an explicit calibration drive.

Odometer Pulse Calibration. Different vehicle models provide signals with different wheel pulses per unit distance. The dead-reckoning software can easily match the wheel pulses with the distance traveled as measured with GPS. Driving a short straight route is normally sufficient to calibrate the wheel pulses. Calibration should be an ongoing process to account for small variations such as those due to changing tires.

Gyroscope Calibration. Concerning the gyroscope, two parameters need to be calibrated: voltage offset and volts per degrees per second.

The first parameter is obtained quickly while standing still. Calibration of the second parameter requires driving multiple curves in both directions.

Figure 8 Temperature effects on gyroscope offset voltage (measured example)
Furthermore, the gyroscope exhibits temperature-dependent characteristics and ageing effects. Temperature dependencies as shown in Figure 8 are managed with a table to record gyroscope offset voltages at different temperatures. The on-going calibration process during normal operation compensates ageing effects.

Test Drives The densely populated areas with high-rise buildings and deep urban canyons in New York City are challenging locations to test the performance of dead-reckoning-enabled GPS receivers. While GPS-only positioning causes significant degradation in accuracy in the midtown and downtown area, the dead-reckoning GPS receiver provides reliable uninterrupted navigation.

Pages | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8

Innovation: GPS Time Transfer
Innovation: Dam Stability
Innovation: Phase Wind-Up Analysis
Innovation: GPS + LORAN-C
Innovation: Searching for Galileo



Click here
view GPS Digital archives
Click here to Subscribe


RSS feeds for this site