Robotics and Embodied AI
Perception, planning and control for robots that learn: frames and kinematics, sensors and perception, state estimation, localization and SLAM, motion planning, feedback control, reinforcement learning and sim-to-real transfer, imitation learning, foundation models and vision-language-action models, and robot safety and deployment.
AI-450 Year 4 3 credits 12 lessons · 4 h 50 min
Offered by HSM University · Artificial Intelligence
This is the American version of this course. See the Canadian version.
What you'll learn
- Describe robot types, the sense-plan-act loop and its timing, ROS 2 architecture and the role of simulation
- Use rotation matrices, quaternions and homogeneous transforms to compute forward kinematics and manage coordinate frames
- Solve inverse kinematics analytically and numerically, analyze Jacobians and singularities, and explain manipulator dynamics
- Model cameras and depth sensors, select sensors for tasks, and build perception pipelines including foundation-model detection
- Fuse sensor data with Kalman, extended and complementary filters, with gating and calibrated uncertainty
- Localize robots with particle filters, build occupancy grids, and explain graph-based SLAM and loop closure
- Plan motions with A*, sampling-based planners and trajectory optimization, and time-parameterize trajectories
- Design PID, impedance, LQR and model predictive controllers and explain whole-body control
- Train control policies with reinforcement learning and close the sim-to-real gap with randomization and teacher-student training
- Learn policies from demonstrations with behavior cloning, DAgger, action chunking and diffusion policies, and evaluate them honestly
- Explain how language, vision-language and vision-language-action models are used in robots, with grounding and their limitations
- Apply robot safety standards, collaborative operating modes and safety layers for learned systems, and plan deployment
Course content
12 lessons · 4 h 50 min total · final exam
- 1 Embodied AI and the robot system 25 min · quiz
- 2 Coordinate frames, transforms and forward kinematics 24 min · quiz
- 3 Inverse kinematics, Jacobians and dynamics 24 min · quiz
- 4 Sensors and robot perception 24 min · quiz
- 5 State estimation and sensor fusion 24 min · quiz
- 6 Localization, mapping and SLAM 24 min · quiz
- 7 Motion planning 24 min · quiz
- 8 Feedback control for robots 24 min · quiz
- 9 Reinforcement learning for control and sim-to-real transfer 24 min · quiz
- 10 Imitation learning from demonstrations 24 min · quiz
- 11 Foundation models for robotics 25 min · quiz
- 12 Safety, standards, deployment and capstone 24 min · quiz
- E Final exam 20 questions
Requirements
Recommended background. You can start any course at any time.
Description
Perception, planning and control for robots that learn: frames and kinematics, sensors and perception, state estimation, localization and SLAM, motion planning, feedback control, reinforcement learning and sim-to-real transfer, imitation learning, foundation models and vision-language-action models, and robot safety and deployment.
Part of the Science stream, typically taken in year 4.