Getting there but not quite there yet

Currently I have the two accelerometers averaging out their data in processing to figure out tilt. There are two problems. First one is it cannot pass the 180 point. When it does, it reverts to its original position. Secondly it is very shaky and has issues across the x axis in 3D. I am seeking out some better math equations. I want this project not to require calibration which would fix some of my problems currently. I am going to speak to my professors over some of the math required to calculate the position.