1 Intro

CS4246: AI Planning
& Decision Making

Semester 1, 2017-2018
Outline
Course administration
Introduction to AI planning & decision making
Decision-Theoretic Agents
2
Course Administration
Teaching staff
Teaching resources
Objective
Syllabus
Assessment overview
3
Teaching Staff
Lecturer: Bryan Low
Email: lowkh@comp.nus.edu.sg
Website: http://www.comp.nus.edu.sg/~lowkh
Office: COM2-02-58
Consultation hours: By appointment
Research interests: Planning under uncertainty, machine learning, multi-

agent systems, robotics
TAs: Teng Tong, Sreejith Balakrishnan

Email: tengtong@comp.nus.edu.sg, sreejithb@gmail.com
Consultation hours: By appointment
4
Teaching Resources: IVLE
https://ivle.nus.edu.sg/
Lesson plan
Lectures, tutorials, homeworks, projects, supplementary materials
Discussion forum
Any questions related to the course should be raised on this forum
Emails to me will be considered public unless otherwise specified
Announcements
Homework and project submissions
5
Objectives
What is CS4246 about?
Introduce foundational concepts and practical implications of AI planning
and decision making
Survey state-of-the-art advancements in theories and applications of

planning and decision making technologies
What will you learn in CS4246?

Understand the main concepts, issues, and challenges in AI planning and
decision making technologies
Apply the technologies in different application domains
Research further into AI planning and decision making
6
Objectives
Who should take CS4246?
Undergraduates and beginning graduate students. Centered towards CS or
by permission.
This module is theory-oriented: Modern AI uses and exploits considerable

math, Bayesian statistics, and algorithms.
Prerequisites
CS1231 Discrete structures

MA2216 or ST2131 or ST2334 Probability and statistics
CS3243 Introduction to artificial intelligence
Linear algebra and calculus (strongly preferred)
7
Main Reference
Russell and Norvig (2010).
Artificial Intelligence: A
Modern Approach (3rd
Edition Important!)
Online Resources, Code,

and ERRATA:
http://aima.cs.berkeley.edu/
Additional Resources
You will also find good tutorials, tools, & publications at
International Conference on Autonomous Agents and Multiagent Systems
(AAMAS) http://www.aamas-conference.org
AAAI Conference on Artificial Intelligence (AAAI) http://www.aaai.org
International Joint Conference on Artificial Intelligence (IJCAI)
http://ijcai.org
Conference on Uncertainty in Artificial Intelligence (UAI)
http://www.auai.org
Neural Information Processing Systems Conference (NIPS) http://nips.cc
International Conference on Machine Learning (ICML) http://icml.cc
International Conference on Automated Planning and Scheduling (ICAPS)
http://www.icaps-conference.org/
9
Syllabus
Decision-theoretic agents
Planning under uncertainty
Planning and decision making in the real world
Planning and learning
Multi-agent decision making

10
Public Holidays
1 Sep (Week 3) Hari Raya
Haji > No lecture
18 & 20 Oct (Week 9)

Deepavali > No tutorials
10 Nov (Week 12) WEF

Annual Meeting of GFCs
2017 > No lecture
11
Assessment Overview
Tests (2) 60%
13 Oct 2017 (Fri, 6:30-8:30pm): During lecture
14 Nov 2017 (Tues, 6:30-8:30pm): Venue TBA
NO make-up
Projects (TBA) presentation on 17 Nov 2017 35%
Graded tutorials & class attendance 5%
No final exam
12
Freedom of Information Rule
Collaboration is acceptable and encouraged.
You must always write the name(s) of your collaborators on
your assignment.
You will be assessed for the parts for which you claim is
your own contribution.
13
On Collaboration
You are free to meet with fellow students(s) and discuss
assignments with them.
Writing on a board or shared piece of paper is acceptable
during the meeting; however, you may not take any written
(electronic or otherwise) record away from the meeting.
After the meeting, do something else for at least a half-hour
before working on the assignment.
This will ensure that you are able to reconstruct what you
learned from the meeting, by yourself.
14
Introduction to AI Planning
& Decision Making
in the Real World
Rational Agents (CS3243 Revisited)
Rational Agent. For each possible percept sequence, a

rational agent should select an action that maximizes its
expected performance, given the evidence provided by the
percept sequence and whatever built-in knowledge the agent
has.
Performance measure. Objective criterion for measuring
success of an agents behavior.
16
Motion Planning of LittleDog
over Complex Terrains
SideKick in Cooperative Games
Intention-Aware Planning
for Autonomous Vehicles
Learning to Plan from
Expert Demonstrations
Planning as Search (CS3243 Revisited)
Search is the process of determining a sequence of actions
from initial state (Outram Rd) to goal (Old Airport Rd).
Goal
Formulate problem.
Abstraction into goal,
states, actions, cost,
transition model
Solve problem. Choose
search strategy (e.g., A*)
to output action sequence
Assumptions of task
environment. Fully
observable, discrete,
deterministic Initial
State 21
Properties of Task
Environment (CS3243 Revisited)
Fully observable (vs. partially observable). An agents
sensors give it access to the complete state of the
environment at each point in time.
Deterministic (vs. stochastic). The next state of the

environment is completely determined by the current state
and the action executed by the agent.
Episodic (vs. sequential). The agents experience is divided

into atomic episodes (each episode consists of the agent
perceiving and then performing a single action), and the
choice of action in each episode does not depend on actions
in past episodes.
22
Properties of Task
Static (vs. dynamic). The environment is unchanged while
an agent is deliberating.
Discrete (vs. continuous). A finite number of distinct states,

percepts, and actions.
Single agent (vs. multi-agent). An agent operating by itself

in an environment.
23
Properties of Task
Task Toy Real
Autonomous vehicle in UTown
environment problems world
Fully observable Yes No Hidden intentions of humans

Deterministic Yes No Wheel slippage
Episodic Yes No Motion constraints
Static Yes No Humans are in motion
Discrete Yes No Continuous states & actions
Single agent Yes No Multiple humans
24
References
RN 2, 3.1-3.4, 3.5.1-3.5.2, 3.6.1-3.6.2
CS3243 lecture notes 1, 2, 3
25

1 Intro

Uploaded by

Copyright:

Available Formats

1 Intro

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 Intro

Uploaded by

Copyright:

Available Formats

CS4246: AI Planning

& Decision Making

Consultation hours: By appointment

Research interests: Planning under uncertainty, machine learning, multi-

TAs: Teng Tong, Sreejith Balakrishnan

Consultation hours: By appointment

Lectures, tutorials, homeworks, projects, supplementary materials

Any questions related to the course should be raised on this forum

Emails to me will be considered public unless otherwise specified

Homework and project submissions

Survey state-of-the-art advancements in theories and applications of

What will you learn in CS4246?

Apply the technologies in different application domains

Research further into AI planning and decision making

This module is theory-oriented: Modern AI uses and exploits considerable

CS1231 Discrete structures

Online Resources, Code,

18 & 20 Oct (Week 9)

10 Nov (Week 12) WEF

13 Oct 2017 (Fri, 6:30-8:30pm): During lecture

14 Nov 2017 (Tues, 6:30-8:30pm): Venue TBA

Projects (TBA) presentation on 17 Nov 2017 35%

Graded tutorials & class attendance 5%

Rational Agent. For each possible percept sequence, a

Deterministic (vs. stochastic). The next state of the

Episodic (vs. sequential). The agents experience is divided

Discrete (vs. continuous). A finite number of distinct states,

Single agent (vs. multi-agent). An agent operating by itself

Fully observable Yes No Hidden intentions of humans

You might also like