Gyu-In Jee

The interior-point method for an optimal treatment of bias in trilateration location

This paper presents a new position-determination estimator for trilateration location. The proposed estimator takes the measurement bias into consideration and improves the location accuracy of a mobile location system. In case that a mobile station (MS) utilizes signals from a set of base stations for its location, the computed location is largely affected by nonline-of-sight (NLOS) error in signal propagation. A constrained optimization method in a three-stage estimation structure is proposed to estimate and eliminate the measurement bias contained in each pseudorange and mainly caused by the NLOS error. A linear observation model of the bias is formulated, and the interior-point optimization technique optimally estimates the bias by introducing a feasible range of the measurement bias. It is demonstrated that the new three-stage estimator successfully computes an accurate location of an MS in a realistic environment setting. The location accuracy of the proposed estimator is analyzed and compared with the existing methods through mathematical formulations and simulations. The proposed estimator efficiently mitigates the effect of a measurement bias and shows that the iterated least square (ILS) accuracy of 118 m [67% distance root-mean-square (DRMS)] can be improved to about 17 m in a typical urban environment

Efficient use of digital road map in various positioning for ITS

There are many R&D improvements on positioning systems for ITS (intelligent transportation systems) adopting GPS, cellular phones or other communication systems. But, a position from any signal is always corrupted to a few meters through several hundreds of meters because of multipath, atmospheric delay, NLOS (non-line-of-sight), low DOP and so on. When the positioning systems are employed for ITS, a digital road map can be used together to display their navigation solutions in most ITS applications. Due to the fact that land-vehicles almost always run on roads, most of CNS (car navigation systems) translate the measured position onto a road. This methodology called map-matching, if it depends on a contaminated position due to white noise and biased error, has not only low accuracy but also the road ambiguity problems in some crossroads. Therefore, this paper presents an efficient use of an advanced map-matching in order to get a more improved accuracy, which estimates a large bias being the main source of errors and corrects a vehicles position. It is composed of a modeling of biased error and filtering by a Kalman filter. We have applied the proposed map-matching to not only GPS navigation but also CDMA location. The proposed approach represents that in addition to its original visual display, an accurate digital road-map can improve the positioning accuracy effectively by correcting the vehicles position.

A satellite selection criterion incorporating the effect of elevation angle in GPS positioning

Abstract When positioning by global positioning systems, geometric dilution of precision is generally used as a satellite selection criterion. While it takes account of only geometric effects, there are other factors causing positioning error, such as ionospheric delay, tropospheric delay, and multipath. In this paper, a new criterion for satellite selection, which considers the effect of these errors, is proposed and the effectiveness of the proposed criterion is shown.

Cancellation of Correlation Side-Peaks for Unambiguous BOC Signal Tracking

Based on a combination of the sub-correlation functions composing the autocorrelation of the binary offset carrier (BOC), we propose a novel cancellation scheme of correlation side-peaks to solve the ambiguity problem in the BOC signal tracking. The proposed scheme is applicable to both generic sine- and cosine-phased BOC signals without requiring any auxiliary signal in the receiver. Avoiding the ambiguity problem, the proposed scheme is confirmed from numerical results to provide a better tracking accuracy than the conventional schemes.

GPS multipath detection based on sequence of successive-time double-differences

Based on a sequence of successive-time double-differences of pseudorange and carrier phase measurements provided by the Global Positioning System (GPS), a channelwise test statistic is proposed. The proposed test statistic is particularly efficient in multipath detection and operationally advantageous: it requires only basic observables, it requires no precomputation of receiver position, it has standard /spl chi//sup 2/-distributions, and it can reduce the detection resolution by increasing the number of successive measurements. To analyze the proposed test statistic, both minimum detectable jump and minimum detectable ramp are derived. An experiment under an intentional multipath environment exhibits the effectiveness of the proposed test statistic.

Compensation of gyroscope errors and GPS/DR integration

In car-navigation, a dead-reckoning (DR) system is required for solving the problem of GPS signal blocking in urban area. The vibrating gyroscope which is widely used to determine the vehicle heading in a DR system, has a low accuracy, hence an error compensation of it is needed. In this paper, an error compensation method for the vibrating gyroscope used in DR is presented. Gyroscope errors can be characterized into two types of error, deterministic and random errors. Each error term is mathematically modeled and an indirect feedback Kalman filter is used for the error compensation. Through laboratory experiment and field-test, it is shown that the error of a vibrating gyroscope can be successfully compensated by the proposed method. Also, as an alternative method for GPS/DR integration in a case that the number of visible GPS satellites is insufficient for the correct determination of a vehicle position, the mixed-measurement algorithm is proposed. It is shown that by using this algorithm, even if the number of visible GPS satellites is 2 or 3, the position error becomes smaller than by using only DR.

GPS/DR Error Estimation for Autonomous Vehicle Localization

Autonomous vehicles require highly reliable navigation capabilities. For example, a lane-following method cannot be applied in an intersection without lanes, and since typical lane detection is performed using a straight-line model, errors can occur when the lateral distance is estimated in curved sections due to a model mismatch. Therefore, this paper proposes a localization method that uses GPS/DR error estimation based on a lane detection method with curved lane models, stop line detection, and curve matching in order to improve the performance during waypoint following procedures. The advantage of using the proposed method is that position information can be provided for autonomous driving through intersections, in sections with sharp curves, and in curved sections following a straight section. The proposed method was applied in autonomous vehicles at an experimental site to evaluate its performance, and the results indicate that the positioning achieved accuracy at the sub-meter level.

A complete GPS/INS integration technique using GPS carrier phase measurements

For its synergistic relationship, the integrated GPS/INS system has been adopted in many navigation systems. In this paper, a complete GPS/INS integration technique utilizing GPS carrier phase measurements is proposed. A measurement model of double-differenced GPS carrier phase measurements is newly derived in order to be used with the INS error model. Also, an algorithm is suggested to resolve the integer ambiguities of GPS carrier phase measurements by employing MS information. From simulation results, performance of the suggested technique is verified.

The stability of the adaptive two-stage Extended Kalman filter

The well-known conventional Kalman filter requires an accurate system model and exact stochastic information. But in a number of situations, the system model has unknown bias, which may degrade the performance of the Kalman filter or may cause the filter to diverge. The effect of unknown bias may be more pronounced on the extended Kalman filter. The two-stage extended Kalman filter (TEKF) with respect to this problem has been receiving considerable attention for a long time. In the case of a random bias, the TEKF assumes that the information of a random bias is known. But the information of a random bias is unknown or partially known in general. To solve this problem, the adaptive two-stage extended Kalman filter (ATEKF) for nonlinear stochastic systems with unknown constant bias or unknown random bias was proposed by Kim and coauthors. This paper analyzes the stability of the ATEKF. To analyze the stability of the ATEKF, this paper shows that firstly the adaptive augmented state extended Kalman filter (ASEKF) is equivalent to the ATEKF and secondly the adaptive ASEKF is uniformly asymptotically stable. The analysis result shows that the upper bound of the error covariance must be appropriately bounded for the filter stability.

An error analysis of GPS compass

In this paper, analytical error characteristics of GPS compass which determines heading and elevation of the vehicle using carrier phase measurements from two antennas is given. The derived result gives clear analysis of error statistics on attitude determination using GPS carrier phase measurement. It shows the accuracy of measured attitude relies on satellite geometry, performance of receiver, baseline length, baseline configuration and nominal attitude of vehicle. It is also shown that to obtain more precise heading, longer baseline along the vehicle is preferred and the accuracy of heading is always better than that of elevation since the vertical measurement uncertainty caused by geometry of satellites affects only on that of elevation. These results can be applied to basic directions for attitude determination: to what direction should the baseline be located to minimize the error, which satellites should be selected to minimize the error. These must be important and useful pre-information in attitude determination problem. An application to real data shows that the predicted error by the proposed method coincides with that of experiments.