Week 0 - 21/7
A
No lecture A in week 0
B
Re-schedule classes to try and accommodate everyone.
Lab
No labs week 0
Week 1 - 28/7
Course Overview
Agent/Planning systems
rUNSWift overview
Aperios
Our Objects/systems
Low/High level vision
Tracking
Behaviours
Wireless
ActuatorControl
Flow Control between objects
Interfaces between objects
Lab (Level 3)
Intro to rUNSWift agent
Subversion
Compiling
Running
Base Station
Week 2 - 4/8
Robot Vision
Colour segmentation
Blobbing
High level concepts
Motion (low level)
Head Control
Forward and Inverse Kinematics (Head and Legs)
Point Projection
Non-Commutable Rotations
Recorded motions
Lab (Level 3)
Ball tracking
Assignment 1 out at end of lab (Tracking & Life Saver)
Assignment 2 out at end of lab (Grab Ball - get to other half of field)
Assignment 3 out at end of lab (1x1 penalty shooter)
Week 3 - 11/8
Tracking
Bayesian Localization I
Tabular representation
Tracking
Bayesian Localization II
Particle Filters
Kalman-Bucy Filters
Lab (Level 3)
Go To Point
Experiments with tracking (odom/obs, bad mean/var)
Week 4 - 18/8
Finish Tracking
Information form Kalman Filters
SLAM
Expectation Maximisation if time
Motion (Walking)
Walking
Normal Walk
Odomentry Calibration
Parameter Optimisation
Lab (Level 3)
Work on Assignment 1
Week 5 - 25/8
Wireless
UDP vs TCP
Behaviours
Hysteresis
Hierarchies
Software Engineering and State
Agent Architectures
Engineered (subsumption)
Logic based (Three layer, planning, golog)
Cognitive (ACT-R, not Golog in this sense)
Reinforcement Learning
Lab (Level 3)
Work on assignment 1
Week 6 - 1/9
Introduction to Search
BFS/DFS
A*
Islands
Further search - heuristics
Mixing Search and Heuristics (Korf)
Heuristics as simplifications/relaxations
Lab (Level 3)
Assignment 1 graded in lab
Week 7 - 8/9
Reinforcemnt Learning
Style of problem
Optimality criteria
Exploration/Exploitation trade-off
Planning
State Space search
Lab (Computer lab)
A*/Islands/Heuristics
Week 8 - 15/9
Markov Decision Processes
Formal definition
Model based updates
Certainty equivalence
Prioritized Sweeping
Model free updates
Q-learning
Adaptive Heuristic Critic
Planning
Plan space planning
Lab (Level 3)
Work on Assignment 3
Week 9 22/9
Applying discrete search to continuous problems
Obstacle expansion
Different representations
Rapidly-Exploring Random Trees
Introduction to Optimization
1D Golden Section Search
Assignment 2 Due at start of lecture.
Lab (Computer lab)
Planning
Mid Semester Break - 29/9
Week 10 - 6/10
Gradient descent
Simplex search
Powell's Method
Gradient Descent
Conjugate Gradient
Quasi-Newton methods
Gradient estimation
Handling noise
Oversampling
Pegasus and pseudo-randomness
Function fitting
Meta-search
Examples
Walk learning as application of optimization
Gradient descent over a stochastic policy in gridworld
Lab (Computer Lab)
Optimization
Week 11 - 13/10
Partially Observable Markov Decision Processes
Belief state Tracking based methods
History methods
Explicit memory methods
Piecewise linear value function
Genetic Algorithms
Shock
Lab
No Lab - use time for review if needed.
Week 12 - 20/10
Adversarial Agents
Minimax search
Stochastic Policies
Best Response Learning
Review
Lab (Level 3)
Assignment 3 graded in lab