Improving GPS Location Availability and Reliability by Using a Suboptimal, Low-Cost MEMS Sensor Set

There are a variety of automotive applications which require vehicle location information with a high degree of availability and reliability. Location information derived from high-sensitivity GPS technology can fulfill this need for a substantially large percentage of time. There are several instances where GPS signals are not available: For example in tunnels or parking garages. In weak signal or multipath environments, the reliability and availability of position derived from GPS receiver might be lower than expected. In such environments, to guarantee availability of position with a desired level of reliability, external inputs, such as MEMS inertial sensors, can be used for location computation. It is well known that GPS and inertial sensors are complementary in their error characteristics. The GPS position solution has long term stability characteristics. In contrast, the error in position computed using inertial sensors in short term is relatively small, but it has unbounded error growth with passage of time. Ideally, GPS/INS integration requires measurements from a 6-Degrees-of-Freedom (DOF) inertial measurement unit (three acceleration and three angular rate measurements). Since the automotive application restricts the motion of the vehicle, a suboptimal sensor set can be used. This paper presents the requirements for a minimum set of sensors needed when the target platform is an automobile. It is shown that when using the automobile platform motion constraints, a full six degree of freedom sensor set is not required to maintain location availability for as long as 2 minutes when using low cost MEMS sensors. A loosely coupled algorithm is presented, which uses this minimum sensor configuration to solve for the location parameters of interest. This loosely coupled algorithm is shown to be insensitive to small mounting misalignments and does not require factory calibration procedure to estimate sensor parameters and alignment. This allows the system to be used right out of the box in an automobile without loss of performance. The performance results of this system are presented in a variety of real world weak signal and multipath environments such as high and low urban canyon, tunnels, and dense foliage areas. A comparison of these results with GPS-only data demonstrates marked improvement in positioning performance and reliability of combined GPS/INS system.