Systems and methods for unsupervised indoor localization are provided. Sensor data obtained from a device carried by a user can be used to simultaneously estimate the indoor location of a user and identify landmarks within the indoor environment based on their signatures. The landmarks can be used to reset the location estimate of the user, and the location estimate of the user can be used to improve the learned location of the landmarks. This recursive process leads to excellent accuracy in indoor localization. Systems and methods for estimating the heading direction of a user are also provided. Sensor data obtained from the user can be used to analyze the forces acting on the user in order to give an accurate heading direction estimate.