Robots That Perceive is an open, notebook-based textbook. Every section is a Jupyter notebook: read the explanation, study the equations, then run the code, change a parameter, and watch the robot behave differently. Everything runs in your browser. There is nothing to install.
The book is organized around one idea. A robot senses its world, reasons about what it sensed, and acts, over and over. Part I builds the tools for each step. Part II applies them to six robots, each adding one new idea.
Part I · Foundations¶
What a robot is, the sense–think–act loop, and how we model states, actions, and sensors.
Probability, sensor models, Bayes’ rule, and making decisions under uncertainty.
Part II · Robots, One Idea at a Time¶
Discrete states, sensor fusion with naive Bayes, and learning sensor models from data.
Markov chains, hidden Markov models, the Bayes filter, and planning with MDPs.
Continuous state, particle-filter localization, and A* path planning.
Poses and kinematics, cameras and image processing, and feedback control.
The bicycle model, pure pursuit, Kalman filtering, and LiDAR occupancy mapping.
3D rotations, multirotor dynamics and PID control, and smooth trajectories.
Who this book is for¶
The book assumes first-year calculus, some linear algebra (matrices and vectors), and basic Python. No previous robotics or probability course is needed. Each section takes 30–60 minutes and ends with Try it exercises and key takeaways. It works for self-study or as the backbone of a one-semester introductory course.