doc: add an introduction to the index page.

doc/index.rst

@@ -1,5 +1,37 @@
+starman: state estimation library
+=================================
+
 .. toctree::
     :maxdepth: 2
+    :hidden:
 
     kalman
     reference
+
+The starman library implements state estimation algorithms for tracking the
+hidden state of a system given noisy observations. Take a look at the
+:doc:`reference` for all he gory details.
+
+Features
+--------
+
+A flexible Kalman filter implementation is provided via the
+:py:class:`starman.KalmanFilter` class. Use a Kalman filter to estimate the
+hidden state of a linear system in the presence of Gaussian noise.
+
+Kalman filters are "online"; they provide estimates of the current state of the
+system given past measurements. If you have all the measurements available to
+you, you can use an "offline" method which estimates the state at each time
+instant given all past *and* future measurements. A Rauch-Tung-Striebel (RTS)
+smoother uses the Kalman filter outputs and recursively computes this "all data"
+estimate. The :py:func:`rts_smooth` function provides a RTS implementation.
+
+See :doc:`kalman` for more details and example code.
+
+Why "starman"?
+--------------
+
+Starman implements the Kalman filter. The Kalman filter was used for trajectory
+estimation in the Apollo spaceflight programme. Starman is thus a blend of
+"star", signifying space, and "Kalman". That and "kalman" was already taken as a
+package name on the PyPI.

doc/reference.rst

@@ -1,5 +1,5 @@
-Reference
-=========
+Programmer's Reference
+======================
 
 .. autoclass:: starman.KalmanFilter
     :members: