kalman_filter
An implimentation of a kalman filter, a tool to estimate values in an uncertain environment with uncertain measurement
The Kalman Filter is used to estimate conditions where we are uncertain about value and we recieve measurements.
A good example is the 1 dimensional Brownian motion, where we have an object a some position X At any moment, the position is uncertain with a gausian distribution. For every time step (of an arbitrary amount) we apply a transition function. In the case of the 1 dimensional movement, we simply flatten the gausian (the object randomly moves in some as-yet-unknown direction) Finally, we udpate the gausian with a measurement (it itself a gausian distribution) which we combine with our prior.
The final process looks like this
- Take some initial position
- Update position
- Take measurement
More complicated examples can take the form of movement with aceleration, speed and position.
This work will be used to hopefully represent position for a drone (to come)