1. Finite State Machines ENGR 110 #7 2016

2. Menu Designing a controller for traffic lights.
What is a Finite State Machine?
Finite State Machines for controllers.
Announcements:

3. Traffic Lights Traffic lights at Salamanca Rd/The Terrace
They are annoying!
How should they work?
Simple – only a three way intersection
Write down a design.

4. Finite State Machine What are FSM's?
A way of thinking about certain kinds of problems
A way of describing/modeling/analysing systems
A way of designing solutions to a wide range of engineering problems.
Example applications
Models of physical devices
Controllers for physical devices
Controllers for NPCs in computer games
Protocols for communications/networking
Regular expression searching in text
Analysing and Designing user interfaces.
Speech recognition ….

5. States and Transitions
Key idea:
If a system has finite set of possible states, then you can list them, and identify each of the possible things that could happen in each state.
Typically:
draw labelled circles for the states
draw labelled arrows for the transitions between them
transition1
transition3
State1
State2
transition2

6. States and Transitions
Eg: modelling a Door:
Possible states: open, closed
Possible transitions: push, pull
(Take system from one state to another ).
, closed&Locked
, turn key
or back to same state)
push
turn key
push
pull
push
open
closed
closed
&
locked
turn
key
pull
pull
turn key

7. States and Transitions
Eg: an ecommerce website:
states: the series of pages in the purchase sequence
transitions: the buttons: buy now, next, back, confirm, back to cart
Useful model when first designing the web pages.
Cart
back to cart
back to cart
back to
cart
buy now
next
next
confirm
Address
Details
Payment
Details
Confirm-
ation
Receipt
back
back

8. Abstraction Typically, a Finite State Machine is an abstraction:
Simplification
Capturing the significant aspects of the behaviour
Leaving out unimportant details
Describes the system at a particular level.
eg: the door:
how far open the door is,
how big the door is,
what kind of key it uses
how hard you have to push to close it. ….
eg the website
no details of the actions on each page

9. Kinds of Finite State Machines.
Lots of different varieties of Finite State Machines.
Describing how a passive system works.
states the system can be in,
actions on the system that change its state.
eg: the door, the website, a DVD player,
Can use the FSM description to
Design the system
Explain the system to users
Construct a simulation of the system ….
open
closed
&
locked
pull
push
turn key
turn
key

10. Kinds of Finite State Machines.
Describing/specifying an active system
states the system can be in
input/signals/sensors that the system may respond to
actions that the system will perform
eg: an autonomous vehicle with sensors
a controller for a robotic toy
an NPC in a video game
a controller for a lift (the assignment) …
begging
patted
start
begging
stop
stop
timer
timer
walking
waiting
turning
round
patted
IR sensor
start walk, start timer(30)
start turning, start timer(10)

11. Kinds of Finite State Machines.
Specifying a pattern or a language
states representing different points in the pattern
characters/words/signals that are the next component of the pattern
eg: Regular expression matching in an editor
(SW|NW|EC)EN3[0-9][0-9]
Sequence of the sounds for different phonemes in a word
Simple command language for an adventure game.
(Fundamental to some algorithms for text and language.)
Specifying a communication protocol
states representing different stages in the communication
message types that are expected or required at each stage

12. ? Controllers Devices to control systems. Controller
Could be hardware, but usually software, might be a microprocessor or a large computer.
Controller gets input from the system
Controller performs actions on the system to change its behaviour
Note: two different systems here:
controller and the controlled system.
Might need to design both, but they are separate issues.
System:
robotic toy
autonomous vehicle
lift
traffic lights
vending machine
animation of NPC
etc.
Controller ?
Signals/sensors
actions

13. Simple Reactive Controllers.
Controller drives the system:
Input: Signal/sensor value from the system triggers an action
Output: Action, reacts to the signal
Same response to a sensor, every time
Very limited
System:
robotic toy
autonomous vehicle
lift
traffic lights
vending machine
animation of NPC
etc.
Signals/sensor values
Controller
Signal1 ⇒ action1
Signal2 ⇒ action2
Signal3 ⇒ action3
Signal4 ⇒ action4
actions

14. FSM Controllers. Controller drives the system: System: Controller
Input Signal/sensor from the system
triggers an action
and changes state (of the controller)
States represent the history of what has gone before to enable different responses to same sensor at different times
Much richer behaviour.
System:
robotic toy
autonomous vehicle
lift
traffic lights
vending machine
animation of NPC
etc.
Signals/sensors
Controller
sensor
action
actions

15. Example: Traffic light controller
Signals/sensor values?
Actions?
States?
what are the different conditions where the controller should act differently?
eg:
should it always change the lights when a car drives up?

16. States for traffic light controller