# Stabilizing method

This document is an advanced document on the algorithms of the stabilizing method.

- 1.It takes the average of the input history, and it caused delayed and slow motion.
- 2.In stabilizing process, the method relies on the fixed time interval to stabilize, but it is not a good choice for low-end PC.

Now, we introduced the damping factor to remedy the first problem. The damping algorithm is following:

$output = {\sum_{i=1}^{n} {{x_i} \over {d^i} } \over {\sum_{i=1}^{n} {{1}\over {d^i}}}}$

d is equal to or greater than 1, and n is the damping window size. The latest input has index 1. If d=1, then the algorithm is just average. If d becomes larger, it means that the latest input becomes more dominant than the older input.

We converted the motion pipeline into an event-driven model. It enables to detection of the moment of arrival of new data, and it makes the stabilization process much more rigorous. In the older method, we should poll every new data so we should rely on the fixed time interval, and it caused some problems. The problems are completely removed by using an event-driven model.

Last modified 5mo ago