Due to the rapid development in semiconductor technology, MEMS (microelectromechanical system)-based sensors have been widely used in commercial and military applications. These sensors are characterized by their small size, low cost, low power consumption. However, MEMS sensors are noted for their low accuracy, limiting the applicability in high precision navigation and control. In this paper, a MEMS gyroscope array is developed to account for the limitation of an individual sensor. The key is to develop a robust M-estimation filter to process the sensor array data in real time so as to provide a more accurate estimate of the angular rate, to render fault tolerance capability, and to facilitate signal quality index for system integration. The robust M-estimation Kalman filter is implemented in a DSP/FPGA platform to account for random bias and random walk. In addition, the M-estimation technique is used to accommodate outliers. Allan variance and FFT method are employed as an analyzing tool to quantify the performance. It is verified that the proposed robust M-estimation filter is capable of suppressing non-Gaussian impulse noise and providing a high-Accuracy angular rate measurement.