Catalog Description: This course introduces fundamental concepts, algorithms, and state-of-the-art methods for probabilistic and learning-based robot perception, inference, and localization, with emphasis on multi-modal sensor fusion for autonomous embodied AI systems such as aerial and ground robots. It covers probabilistic state estimation and their extension to modern SLAM frameworks. The course also addresses learning-based predictive models for robotic inference, including neural networks and Gaussian Processes for uncertainty-aware prediction, motion forecasting, planning, and control. A laboratory component includes hands-on experiments and a final project requirement.

Units: 4

Also Offered As: EECS 216

Course Objectives:

Students will learn to: 

• Model and fuse multi-modal sensor data for autonomous robotic systems

• Predict future robot states and trajectories using probabilistic and learning-based methods

• Leverage predictive models for navigation, perception-driven autonomy, and minor planning/control

• Reason about autonomous system behavior under uncertainty Implement probabilistic filters (Bayes, Kalman, nonlinear) and learning-based predictive models (neural networks, JEPA, Gaussian Processes)

• Apply theory to real or simulated robots through hands-on labs and projects

• Employ prediction for planning and control

• Reason about system with autonomous capabilities

Prerequisites: Students are expected to have taken EECS C106A / BioE C106A / ME C106A / EECS C206A or an equivalent course. A strong programming background, knowledge of Python and Matlab, and some coursework in feedback controls and dynamical systems or preliminary knowledge of probability are also useful.

Grading Basis: Student Option

Final Exam Status: No

---

Links:

• EECS Course Archives (CalNet authentication required)
• CS enrollment policies
• EE enrollment policies
• HKN course maps