Robotic Systems
A second course in robotics that moves from single manipulators to complete robotic systems. After a fast review of rigid-body motion and kinematics, it covers mobile robots and sampling-based motion planning (RRT and related methods), and the Lie-group view of orientations and poses used for pose errors and trajectories on SO(3) and SE(3).
The middle part treats state estimation: Bayesian filtering, the Kalman filter and its nonlinear variants (EKF, UKF), and EKF-SLAM. The final part turns to optimization-based control — inverse kinematics and inverse dynamics as quadratic programs, task-space and null-space control, multi-limb and bimanual robots, contacts and friction cones — and ends with how legged robots walk.
- Lec. 1Introduction to Robotics
- Lec. 2Mobile Robots & Motion Planning
- Lec. 3Introduction to Matrix Lie Groups
- Lec. 4Special Euclidean Group & Trajectory Generation
- Lec. 5Kalman Filters
- Lec. 6Non-linear Filters & SLAM
- Lec. 7Introduction to Optimization-based Control
- Lec. 8Optimization-based Control
- Lec. 9Whole-Body Instantaneous Controllers
- Lec. 10Walking Robots
Slides for each lecture, grouped by teaching week.
- Overview of robotic systems and joints
- Notation
- Rigid Body Motions
- Screws/Twists (high-level intuition)
- Jacobians
- FK
- IK (briefly)
- Integrators
- URDF
- Mobile robots
- Motion Planning
- RRT and related
- Orientation Representations
- Quaternions (without the attitude Jacobian)
- Lie Group/Screw Theory/Tangent spaces (SO(3) + SE(3))
- More Lie Group
- Pose errors, rotation errors
- Splines/Trajectories in SO(3)/SE(3)
- Examples
- Probabilities Primer
- Bayesian Inference
- Bayesian Filtering and Estimation
- Motion & Observation Models
- Kalman Filter
- EKF and UKF
- The SLAM Problem
- EKF SLAM
- Inverse Kinematics as an optimization
- Whole Body Inverse Kinematics (QPs)
- Task-space control with Jacobian pseudo-inverse
- Decoupling position and orientation control
- Manipulator Equation and Dynamics
- Force control in joint space
- Force control in task space
- Task-space inertia
- Task-space robust control of manipulators
- Null-space control
- Inverse Kinematics of multi-limb robots
- Kinematic "Balancing" of a biped robot
- Controlling a bimanual robot
- Static balancing on legs
- Whole Body Inverse Dynamics (QPs)
- Controlling with contacts
- Legged Robots
- Contact Dynamics/Friction Cones
- Finite State Machines/Behavior Trees (briefly)
- How do robots walk? (pipeline)
- Walking pattern generation
- Footstep planning
- Walking state machine
- Support Polygon
- Swing foot interpolation