Automated Planning
1. Course Introduction
Jonas Kvarnström
Automated Planning Group
Department of Computer and Information Science
Linköping University
jonas.kvarnstrom@liu.se – 2015

Lecturer:
 Jonas Kvarnström (jonkv@ida.liu.se)
 Computer Science (C program) in Linköping 1992-1996
 PhD, now assistant professor (universitetslektor)
 Leader of the Automated Planning group

Lab Assistant:
 Mikael Nilsson (mikni) – Ph.D. student in planning

Administration:
 Anna Grabska Eklund (annek) – LADOK, …
2
jonkv@ida
About Us…
Please interrupt!
Questions and comments are welcome – start a dialog!
3
jonkv@ida
About the Lectures…
Prerequisites
Basic knowledge and understanding of
data structures and algorithms
as well as logic and discrete mathematics.
Knowledge and understanding of
basic artificial intelligence techniques and concepts,
including search, heuristics and the A* search algorithm.
Essentially no planning knowledge required!
 We will introduce planning concepts
 Then we will quickly go much deeper
5
jonkv@ida
Prerequisites vs. Contents

Today's lecture:
 Introduction to the topic
▪ What is planning?
▪ Where can we use planning?
 Introduction to the course
▪ Contents
▪ Examination
▪ Timetables
 A
▪
▪
▪
closer look at automated planning
Planning vs. reacting
Domain-specific vs. domain-independent
Classical planning
6
jonkv@ida
Today's Lecture
Planning and its Applications
One way of defining planning:
Using knowledge about the world,
including possible actions and their results,
to decide what to do and when
in order to achieve an objective,
before you actually start doing it
More about this definition later…
9
Some applications are simple, well-structured, almost toy problems
Simple structure
General Knowledge
Single agent acting
Specific Knowledge
 There are pegs and disks
 Three pegs, 7 disks
 A disk can be on a peg
 All disks currently on peg 2,
in order of increasing size
 One disk can be above another
 Actions:
Move topmost disk from x to y,
without placing larger disks
on smaller disks
Objective
 All disks on the third peg,
in order of increasing size
jonkv@ida
Towers of Hanoi
10
Some are larger, still well-structured

What you can do:
 Put package p in truck t
 Drive to location
 Deliver p from t

Objective: Deliver 20,000
packages (efficiently!)
 Which packages in which truck,
in which order?
 Which roads to use?
jonkv@ida
Logistics

Classical robot example:
Shakey (1969)
 Available actions:
▪ Moving to another location
▪ Turning light switches on and off
▪ Opening and closing doors
▪ Pushing movable objects around
▪ …
 Goals:
▪ Be in room 4 with objects A,B,C
▪ http://www.youtube.com/watch?v=qXdn6ynwpiI
11
jonkv@ida
Shakey

Tall buildings, multiple elevators
 How to serve people as efficiently as possible?

Schindler Miconic 10 system
 Enter destination before you board
 A plan is generated, determining
which elevator goes to which floor,
and in which order
 Saves time!
 Planning  3 elevators could serve as much traffic
as 5 elevators with earlier algorithms
12
jonkv@ida
Miconic 10 Elevators

Xerox: Experimental reconfigurable modular printers
 Prototype: 170 individually controlled modules, 4 print engines
 Objective: Finish each print job as quickly as possible
13
jonkv@ida
Xerox Printers

Bending sheet metal
 Initially:
There is a flat sheet of metal
 Objective:
The sheet should be in specific shape
 Constraints: The piece must not collide with anything
when moved!
 An optimized plan for bending the sheet
saves a lot of time = money!
14
jonkv@ida
Bending Sheet Metal

Competition: autonomous cars drive 212 km off-road

Requires path planning
 Deciding how to get from one point to another, given:
▪ Speed limits
▪ Constraints on how you can move (turn radius, …)
▪ A map – that may not always be correct
▪ http://www.youtube.com/watch?v=M2AcMnfzpNg 2:00, 4:00
15
jonkv@ida
DARPA Grand Challenge 2005

Competition: 96 km in an urban area (air force base)
 Must follow all traffic regulations, drive around obstacles,
merge with other traffic, …
16
jonkv@ida
DARPA Urban Challenge 2007

SIADEX
 Decision support system for designing forest fire fighting plans
 Needs to consider allocation of limited resources
 Plans must be developed in cooperation with humans – people may die!
17
jonkv@ida
SIADEX

”Remote Agent” on Deep Space 1 spacecraft
 Controlled the spacecraft autonomously for 2 days
 Correctly handled simulated failures
 Rapid response to failures may be crucial to survival!
18
jonkv@ida
NASA Remote Agent

Earth Observing-1 Mission
 Satellite in low earth orbit
▪ Can only communicate 8 x 10 minutes/day
▪ Operates for long periods without supervision
 CASPER software:
Continuous Activity Scheduling, Planning, Execution and Replanning
 Dramatically increases science returns
▪ Interesting events are analyzed
(volcanic eruptions, …)
▪ Targets to view are planned
depending on previous observations
▪ Plans downlink activities:
Limited bandwidth
 http://ase.jpl.nasa.gov/
19
jonkv@ida
NASA Earth Observing-1 Mission

20
Rapid generation of activity plans for Mars Rovers
 Primary platform for creating daily activity plans for Spirit and Opportunity
 Mixed-initiative tool: Human in the loop
jonkv@ida
NASA Mapgen

And why should computers create plans?
 Manual planning can be boring and inefficient
 Automated planning may create higher quality plans
▪ Software can systematically optimize,
can investigate millions of alternatives
 Automated planning can be applied where the agent is
▪ Satellites cannot always communicate with ground operators
▪ Spacecraft or robots on other planets
may be hours away by radio
21
jonkv@ida
Why should Computers Plan?
23
Course Contents
Theory
How do we model
and specify planning problems?
How do planning algorithms work?
How do they relate to and benefit from
different plan structures?
How can planners benefit from
our own deeper domain knowledge?
Practice
Practical experience in
modeling / solving planning problems
using well-known planners
 Deeper understanding of
abilities and limitations
How can we handle uncertainty?
Written Exam
Demos, hand-ins
jonkv@ida
Course Contents

Preliminary lecture order:
 Introduction
 Modeling classical planning domains
 A formal model for classical planning; forward-chaining search
 Forward-chaining heuristics: Intro, relaxations, delete relaxation, …
 Forward-chaining heuristics: hm, hadd, alternative search methods
 Forward-chaining heuristics: Landmarks, Pattern databases
 Domain-configurable planning: Planning with control formulas
 Domain-configurable planning: Hierarchical Task Networks
 Regression planning; planning graphs; extracting heuristics from PGs
 Partial-order planning
 Propositional satisfiability planning
 Path planning
 Markov Decision Processes for probabilistic planning (1; 2)
24
jonkv@ida
Lectures

Main course book
 Reading instructions
are on the web
25
jonkv@ida
Course Book

Lectures vs. Course Book
 Some overlap, but…
26
Lectures:
Overviews
Intuitions
Details suitable for slides
Complementary:
Use both!
(The exam will…)
Book:
Different overviews, intuitions
Additional content and details
Larger examples
Exercises, questions
jonkv@ida
Lectures vs. Course Book

27
Rapid progress in planning research!
 Six labs based on state of the art research prototypes
 Dozens of planners are available
▪ Some ”recommended”, others available as a bonus
Sequential satisficing
acoplan
acoplan2
arvand
brt
cbp
cbp2
cpt4
dae_yahsp
fd-autotune-1
fd-autotune-2
fdss-1
fdss-2
forkuniform
lama-2008
lama-2011
lamar
lprpgp
madagascar
madagascar-p
popf2
probe
randward
roamer
satplanlm-c
sharaabi
yahsp2
yahsp2-mt
Seq. sat. multi-core
acoplan
arvandherd
ayalsoplan
madagascar
madagascar-p
phsff
roamer-p
yahsp2-mt
Seq. optimizing
bjolp
cpt4
fd-autotune
fdss-1
fdss-2
forkinit
gamer
iforkinit
lmcut
lmfork
merge-and-shrink
selmax
Temporal satisficing
cpt4
dae_yahsp
lmtd
popf2
sharaabi
tlp-gp
yahsp2
yahsp2-mt
Older planners
IPP
FF
Specialized planners
SHOP2
jonkv@ida
Labs
Classical planning
1.


Construct a simple planning domain
for emergency service assistance
Investigate properties of several planners
Classical planning
2.



Extensions – learn more about modeling
Use action costs to model plan quality
Test optimal planners
Planning for multiple agents
3.


Using sequential planners – what happens?
Using concurrent planners – what do you gain? How do you model?
28
jonkv@ida
Lab Overview
Hierarchical Task Networks
4.


Defining tasks to perform instead of goals to achieve
Very different modeling task!
Planning with Control Formulas
5.

Using our own domain knowledge to guide a search algorithm
Motion Planning with OMPL
6.

Test a variety of motion planning techniques
using the Open Motion Planning Library
29
jonkv@ida
Lab Overview (2)

Work by yourselves or in pairs
 Working in pairs  must work together!
 Register in WebReg – deadline Tuesday 2015-03-31

If you have a problem:
 First try to solve it yourselves
 Then ask us! Without feedback we can’t help you!

Be at the scheduled labs
 Even though you can work at home!

Work outside the scheduled labs
30
jonkv@ida
Labs

Intended schedule:
 150401 – Finish lab 1 (Classical 1)

Lab assistants are available:
 150416 – Finish lab 2 (Classical 2)
 During scheduled lab hours
 150424 – Finish lab 3 (Concurrent)
 By e-mail,
 150430 – Finish lab 4 (HTN)
 150512 – Finish lab 5 (TLPlan)
 150521 – Finish lab 6 (OMPL)
 150526 – Final lab session

31
Absolute deadlines:
 1506xx – Bonus demo session
in the re-exam period
(date announced later)
 150605 – Written exam
until the demo session 1506xx
 (That's all!)
jonkv@ida
Schedule
32
General policy:
You can always take an exam at least three times per year
 For this course, you can also take the exam:
▪ In the August 2015 re-exam period
▪ In the October 2015 re-exam period
General policy: For all IDA courses having computer lab assignments
there will be one deadline during or at the end of the course.
If you fail to make the deadline, you must retake the possibly new lab course
the next time the course is given.
 For this course, two bonus demo sessions:
▪ In the August 2015 re-exam period
▪ In the October 2015 re-exam period
jonkv@ida
Additional Opportunities
33
jonkv@ida
Strict Deadlines!
34
Do you need money from CSN?
Finish in time, study for the exam!
You can not get points outside of the specified dates
jonkv@ida
Strict Deadlines! 2
35
Will you be leaving the country?
Study for the exam!
You can not take an exam without being here
jonkv@ida
Strict Deadlines! 3
TDDD48 Automated Planning
Questions?