Optimization of Kalman Filter on Accelerometer Data for Automotive Safety Applications

The ever-increasing amalgamation of electronics with the automotive industry in the past decade has seen an integration of various sensors like temperature sensors, RPM sensors, wheel speed sensors, etc. on a vehicle. These sensors have enabled a deep insight into vehicle behavior and a good perception of the operating conditions of the vehicle. The accelerometer is one such sensor, the advancement in the semiconductor industry has bred accelerometer sensors in a MEMs form, which is very cost-effective and also facilitates easy integration because of the microform factor. Moreover, As dictated by AIS 140 norms the Telematics ECUs must have a Triaxial accelerometer & Triaxial Gyro sensor integrated inside them. The data from these MEMs accelerometer and gyro sensors can be used to have a better insight on vehicle dynamics like cabin vibrations, Suspension performance, and External factors like road profile, etc., This data can also be used for safety applications like impact detection, Harsh acceleration (HA) and Harsh braking (HB) detection. One of the major drawbacks of MEMs accelerometers is they are prone to high noises in data. This paper illustrates the filtering of accelerometer data and compares various filters based on the type of application. The major focus of this paper will be on safety-related applications and the Kalman filter due to its feasibility for these applications and the dynamic nature of this filter. Wherein we compare the effect of the process noise covariance (Q) and sensor noise covariance (R) on the Kalman filter and optimize the filter for HA, HB, and Crash detection.