Research On Attitude Measurement Technology Of Robot Based On Multi Sensor Information Fusion

In robot navigation and motion control, acquiring motion attitude in real-time is one of the indispensable and important steps, especially in stable attitude control of self-balancing mobile robot. When implementing intricate action and directional movement, the self-balancing mobile robot needs perceive the change of three degrees of freedom in yaw, pitch and roll at any moment, and adjust its space attitude according to the measurement results, thus keep stable movement, so research on the real-time measurement technology of robot motion attitude has very important significance.Firstly, a robot attitude measurement system based on inertial navigation technology with low-cost MEMS sensor is designed. The hardware of system is mainly composed of five functional modules of data acquisition and processing, MEMS sensor, data storage, voltage conversion and serial communication; while the software mainly includes the driver design of system hardware and the implementation of data fusion algorithm. In attitude measurement, the host controller of system STM32F405RG collects angular velocity and acceleration data from inertial measurement unit MPU6000through SPI bus, meanwhile collects the geomagnetic field data from digital compass HMC5883L through IIC bus, then calculates the current robot attitude with data fusion algorithm, and sends the updated robot attitude to PC through serial with real-time display.For updating of motion attitude in real-time, the paper has proposed two kinds of data fusion algorithms based on quaternion attitude:gradient descent and complementary filtering, and has finished the verification and comparison of these two algorithms in attitude measurement system designed independently. The gyroscope focuses on dynamic performance, while the accelerometer and digital compass focuses on static performance, so the general idea of data fusion algorithm is to implement the high speed integration of gyroscope’s angular velocity, simultaneously conduct the low speed fusion among the accelerometer’s gravity field data, the digital compass’s geomagnetic field data and the results of angular velocity integration, thereby it not only collects the drift error of gyroscope, but also implements the complement of three kinds of sensor in combined application.To compensate the defect of inertial navigation technology that measurement accuracy descends because of the accumulation of random errors of inertial device, another method of attitude measurement based on binocular stereo vision has been studied in the paper, and the main purpose of research is to implement the combined measurement of robot motion attitude with inertial navigation technology and binocular stereo vision. In addition to measure three-dimensional attitude, binocular stereo vision also can be used to spatial localization, so it’s widely used in target detection, tracking and location. In experiment, the paper firstly uses the same two cameras to capture16groups of left and right images, then calibrates camera parameters with matlab, finally uses computer vision library (OpenCV) to finish image rectification and matching.