IMU Senosr fusion algorithm of gyro and accelerometer during acceleration of vehicle

Hello so we have a car with 10 IMUs mounted at different locations of the car and the idea is the measure the centripetal force at each location of the car and compare it to each other. We also need the gyro data as well. So we need the sensor data to have less errors and drift so that we can compare them easily. As you all know, the gyros drift over time which makes calibration a little bit tough. I decided to apply different sensor fusion algorithm like the complementary filter, but I later realised that this only work when the car is not accelerating. Once the the is accelerating then the complementary filter does not output the right values which makes sense.

Now my question is, is there any sensor fusion algorithm which works whiles the car is accelerating at a top speed. My sensor has only 6 DoF, ie. gyroscope and accelerometer.

I will be happy if anyone could let me know of any algorithm which works for accelerating objects. Also, I have the possibility to combine the sensor data with GPS signal from the car but I don't know how it is done.

I will be happy with multiple ideas, algorithms, books, papers and journal.

In general, as you said: "gyros drift over time". The way to compensate for this is:

1. Properly calibrate your sensor. This involves making sure the sensor produces the correct output over a range of rotation rates and over a range of temperatures. The drift should be a random walk. So will be zero mean unless there is a scale error as the chip warms up.

2. Periodically zero the gyro. When you can be certain that your vehicle is not moving, zero out the gyro. i.e. if the gyro is telling you that the vehicle is rotating when you know it is not, then there is some bias that you should subtract out.

Accelerometers don't drift like gyros, but they do need calibration too. They should be calibrated for scale and offset in each axis.