Textbook in PDF format

A key aspect of robotics today is estimating the state (e.g., position and orientation) of a robot, based on noisy sensor data. This book targets students and practitioners of robotics by presenting classical state estimation methods (e.g., the Kalman filter) but also important modern topics such as batch estimation, Bayes filter, sigmapoint and particle filters, robust estimation for outlier rejection, and continuous-time trajectory estimation and its connection to Gaussian-process regression. Since most robots operate in a three-dimensional world, common sensor models (e.g., camera, laser rangefinder) are provided followed by practical advice on how to carry out state estimation for rotational state variables. The book covers robotic applications such as point-cloud alignment, pose-graph relaxation, bundle adjustment, and simultaneous localization and mapping. Highlights of this expanded second edition include a new chapter on variational inference, a new section on inertial navigation, more introductory material on probability, and a primer on matrix calculus.
Introduction
Estimation Machinery
Primer on Probability Theory
Linear-Gaussian Estimation
Nonlinear Non-Gaussian Estimation
Handling Nonidealities in Estimation
Variational Inference
Three-Dimensional Machinery
Primer on Three-Dimensional Geometry
Matrix Lie Groups
Applications
Pose Estimation Problems
Pose-and-Point Estimation Problems
Continuous-Time Estimation