Performance Assessment of An MPU-6000 IMU for Low-Cost Ground Vehicle Navigation
Rodrigo Gonzalez, GridTICs, National University of Technology, Argentina; Paolo Dabove, Department of Environmental, Land and Infrastructure Engineering (DIATI), Politecnico di Torino, Italy
Location: Big Sur
Alternate Number 1
The MPU-6000 inertial measurement unit (IMU) from InverSense Inc. is a low-cost MEMS sensor used in several applications. Its price is around USD 6. This particular IMU and its derivatives (MPU-6050, MPU-9250) can be found in different mass-market products, from wearables, Internet-of-Thing applications, and flight control systems for multicopters, among others. In particular, the MPU-6000 has been chosen to be part of some open hardware autopilots as the Pixhawk and the Beagle Bone Blue flight controllers, and other commercial autopilots as the Parrot Bebop as well.
The MPU-6000 manufacturer provides a datasheet specifying typical noises for a MEMS IMU, but usually this information is a generalization for a production batch. Commonly, a more detailed profile from a specific unit is needed in order to calculate more precise estimations of position, velocity and attitude (PVA) in the context of an integrated navigation system.
Although this mass-market IMU is used extensively, a detailed performance analysis of this inertial sensor is not available in the previous literature. Therefore, it is not clear which could be the field of action of this sensor and in which applications can be effectively used.
In this work a deep examination of the MPU-6000 IMU is provided to verify if this sensor can be used as part of a low-cost navigation system for ground vehicles. The steps to analyze the behaviour of the MPU-6000 inertial sensors is divided in two phases: static and kinematic analyses.
Firstly, an MPU-6000 IMU unit is mounted in a levelled non-magnetic plate and logged during 24 hours to subsequently apply the Allan variance method on these data. The Allan variance is a well-known technique that is used to identify and to quantify inertial sensors' stochastic noises, as quantization noise, angle random walk, and bias instability, among others. The level of these noises is estimated graphically by plotting a log-log curve. Additionally, a six-position static test is carried out to detect systematic errors in this MPU-6000 unit. Besides, two additional MEMS IMU of superior quality are analyzed applying the same procedure described earlier just for the purpose of comparison. These IMUs are one Microstrain 3DM-GX3-35 by Lord, and one Ekinox-D by SGB Systems. They can be considered as industrial-grade and tactical-grade IMUs, respectively.
In a second stage, these three IMUs are mounted on an automobile along with two GNSS sensors, one U-blox M8 receiver and the internal Ekinox-D GNSS receiver. Measurements from all sensors are registered while the vehicle is driven through the streets of Turin City. Several GNSS interference scenarios are forced in this urban trajectory, from crossing a tunnel and heavy leafy foliage to the urban canyon effect, in order to later analyze the performance of an integrated navigation system (INS/GNSS). Finally, three loosely-coupled INS/GNSS systems are created combining the three IMUs and the GNSS receivers, namely, MPU-6000/U-blox, Microstrain/U-blox, and Ekinox/Ekinox.
In addition, a reference data set is made by processing the Ekinox INS/GNSS measurements with tightly-coupled integration. GNSS data is corrected with the Politecnico di Torino permanent station as a master GNSS station, composed by a multi-constellation and a multi-frequency receiver. This is the best possible PVA solution that can be obtained by using effectively all available hardware and software resources. The purpose of the reference dataset is to quantify the performances of the three proposed INS/GNSS systems.
Preliminary results have suggested that the performance of the three INS/GNSS systems are close in PVA. Consequently, an MPU-6000 IMU can be seriously considered as a navigation sensor for low-cost ground vehicle navigation.