Implementation of a simple Histogram Filter for robot localization in a one-dimensional discrete grid world.
The task is seen as a typical filtering problem of a HMM where:
- hidden states: position of the agent on the 1D grid world
- evidence: sensor measurements and the goal is to predict the categorical probability distribution over the states at a point t given all past evidence from 1:t and an initial prior distribution.
See Artificial Intelligence: A Modern Approach - 3rd Edition (Russel, Norvig) Chapter 15.2.1 Filtering and prediction.
The file requirements.txt contains all the libraries needed.
In addition, the library termplotlib used for plotting graphs in the console requires gnuplot. This link refers to the installation on https://wiki.ubuntuusers.de/Gnuplot/.
We use black for formatting, mypy for static typing and flake for linting. The development dependencies are listed under the requirements.txt file.
In order to enforce these, we are using https://pre-commit.com/ for which one installs the git-hooks with pre-commit install.