Title: A New Gravity Absorption Modeling for GPS/RISS in Land Vehicle
Author(s): Jungbeom Kim, Younsil Kim, Heekwon No, Minho Kang, Byungwoon Park, Changdon Kee
Published in: Proceedings of the 30th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2017)
September 25 - 29, 2017
Oregon Convention Center
Portland, Oregon
Pages: 1904 - 1927
Cite this article: Kim, Jungbeom, Kim, Younsil, No, Heekwon, Kang, Minho, Park, Byungwoon, Kee, Changdon, "A New Gravity Absorption Modeling for GPS/RISS in Land Vehicle," Proceedings of the 30th International Technical Meeting of The Satellite Division of the Institute of Navigation (ION GNSS+ 2017), Portland, Oregon, September 2017, pp. 1904-1927.
Full Paper: ION Members/Non-Members: 1 Download Credit
Sign In
Abstract: A GPS/INS Integration system is widely used for navigation of UAV and developing intelligent land vehicle. Normally, the GPS/INS system consists of three-axis gyroscopes, three-axis accelerometers and GPS. It is the most popular method for navigation of moving objects and it considers 15 states such as 3D position, velocity, attitude, accelerometer bias and gyroscope bias. It is a little complicated and it needs hard computational load in small computing devices such as embedded processor to estimate all 15 states. However, with the assumption that the land vehicle mostly stays in the horizontal plane and growing trend that cost reduction of the GPS/INS system, a reduced inertial sensor system (RISS) has been proposed and many researches have been conducted using the GPS/RISS integration system. Among various GPS/RISS methods, 2D GPS/DR is the most general method. It consists of GPS, odometer and one axis gyroscope to measure heading information. An odometer is an instrument that indicates distance travelled by a vehicle and it is based on number of wheel’s rotation. In 2D GPS/DR, it is main sensor like accelerometers in GPS/INS system because the odometer is not affected by influence of gravity. Thus, it is possible to ignore the existence of gravity and it makes the navigation system much more simple rather than above GPS/INS. By the way, it has also disadvantages and error factors. First of all, data sampling rate and resolution which we can easily and legally acquire by using OBDII are insufficient to calculate accurate position like 2 Hz rate and 1km/h resolution. Furthermore, there are sideslip angle, changing pressure of tires and other factors that can make error. Therefore, we have suggested different GPS/RISS with new state for Gravity Absorption. In this study, RISS consists of single-axis gyroscope for heading information and two-axis accelerometers pointing towards the forward and transverse directions. The odometer is not main sensor for navigation any longer, it is just used as additional sensor to provide velocity information. In system model, it is assumed that roll and pitch angles of the land vehicle are zeros. Therefore, this GPS/RISS system operates well for land vehicle simply moving at a constant velocity in the flat road where no roll and pitch angle. However, when the land vehicle drives diverse maneuvers such as rapid acceleration and turning, roll and pitch angle cannot be zeros no longer. Also, it is quite difficult to find flat road in reality. When roll and pitch angle exist, two-axis accelerometers in RISS must be influenced by gravity and GPS/RISS navigation solution can be contaminated. Thus, this paper proposes a new mitigation method of gravity influence for the GPS/RISS system in land vehicle without estimation of roll and pitch angle directly. In other words, a new gravity absorption modeling is suggested to alleviate gravity forces in two-axis accelerometers due to roll and pitch angles. The gravity influences or forces on each axis of the accelerometer are modeled as 1st Gauss-Markov process. They are called, gravity state. To carry out modeling gravity state as 1st Gauss-Markov process, several tests have been conducted in real road to survey environment of real road. After that, we determined time constant (Tau) and noise factor (Q) in 1st Gauss-Markov process modeling based on the previous test results. Then, gravity states for two accelerometers are added to GPS/RISS and can be estimated with other states such as position, velocity, heading information simultaneously. Because they are related the other states, they absorbs negative effects due to gravity force on the other states. We have conducted several simulation and real road test to verify its performance. In the simulation and test results, we have confirmed its positive effect on all states. Furthermore, we have carried out analysis about performance with various time constant. The gravity state is on acceleration domain. However, noise of gravity state due to time constant could be zero on position domain by integration. Therefore, we have found how to model gravity state properly. Then, Gravity state modeling have been improved and position result have been also improved.