CP260-2024: Robotic Perception

Syllabus

Mathematical Model of IMU
- IMU Noise Model, IMU Pre-integration, EKF
Micro-Lie Theory
- Exponential and Logarithmic Maps, Adjoints, Derivatives and Jacobians
Localization & Mapping
- Factor Graphs, Marginalization, Variable Elimination, Non-Linear Least Square
Lidar
- Physics and Architecture, Operation Principles, Scan Matching, Occupancy Grids, Likelihood maps, point-to-point, point-to-plane, KD trees, Iterative Closest Point, Ground Plane Removal
Deep Neural Networks Fundamentals
- Feedforward Networks, Fully connected Networks, Multi-Layer Perceptron, Convolutional Neural Network, Stochastic Gradient Descent, Regularization, Maxpooling, Backpropagation, Skip Connections
3D Point Cloud Processing
- Object Pose Estimation, Object Segmentation, RANSAC, Hough Transforms, Point Pair Features, Normals and Curvature, Deep Learning Methods, GPU Acceleration, Handling Occlusions, Gumbel-Softmax, Pose Symmetries, Mean-Shift clustering
Computer Vision Fundamentals
- Pin-Hole Camera Model and Projections, Homography, PnP, Object Detection, Classification, Tracking, Stereo Processing, Optical flow, Features - Classical and Machine Learning, VIO, Bundle adjustment, Visual SLAM
3D Representations
- Radiance Fields, Gaussian Splatting
Multi-Modal Perception
- Vision + Language, Gestures, Foundational Models