An Introduction To Terrarium

1
An Introduction To Terrarium

• An Ex-Students Perspective

2
Terrarium The Basics

• Real-time game of imperfect information
• You control a bug that lives within a
  Terrarium
• Current information about the environment is
  restricted to the area immediately around and in
  front of your bug
• Lots of bugs can inhabit the same Terrarium at
  once

3
Virus vs. Terrarium
Virus Terrarium
Turn-based Real-time
Perfect information Imperfect information
Interact using an evaluation function Interact using rule based DoAction()
2-player Multi-player
Short games speed essential Longer games speed less vital
4
Virus vs. Terrarium continued

• Those who found Virus evaluation function based
  approach confusing or counter-intuitive may find
  Terrariums rule based approach easier to use
• If two Virus players didnt use random numbers at
  all they would always play the same game
  deterministic
• If two Terrarium players dont use random numbers
  at all there are still random elements outside of
  their control non-deterministic
• Less tuning and optimisation of variables
• More strategy and planning more scope for
  creativity (but only in my opinion!)
• I personally enjoyed Terrarium more than Virus

5
Whats a Terrarium?

• An area surrounded by obstacles within which bugs
  can move and perform actions
• Terrarium Academic includes bugs, rocks
  (obstacles), plants and goals
• Objects are solid
• Objects have different sizes

6
(No Transcript)
7
Terrarium Bugs

• Bugs are introduced into a Terrarium and perform
  actions based upon their code
• Code (should be!) arranged so that the bug reacts
  to its surroundings
• A set of pre-defined methods and variables are
  available to write bug code
• API should be helpful and easy-to-use
• Tool-tip text for all applicable methods and
  variables

8
Bugs continued

• Your bug (and all bugs) have certain attributes
  you can access, e.g.
• Health (Vital!)
• Score (Important!)
• Position
• Speed
• Direction
• Kills
• And others

9
Ticks And DoAction()

• A game of Terrarium takes place in real-time, but
  is still divided into turns
• A turn is referred to within Terrarium as a
  tick
• Each game is comprised of many ticks
• Every tick, each bugs DoAction() method is
  called
• This method is where you write you code
• It tells your bug what to do for the next tick
• Where to move, what to eat, what to attack etc.

10
Ticks And DoAction() continued

• Within your DoAction() function you must write
  code that looks at whats going on around your
  bug and what condition your bug is in, then makes
  a choice about what to do next
• Your method is called within it you call other
  methods that will make your bug perform actions
  like eating and fighting
• You can write your own methods to call from
  DoAction(), and you can use several built-in ones
• Make choices based on the environment and the
  variables that describe your bugs current
  condition

11
Whats the Point?

• The aim in Terrarium (the game not the
  coursework!) is to get a higher score than all
  the other bugs in the Terrarium
• Score points in two ways
• Kill opponents
• Collect points from Goals
• Goals can be hard to reach, difficult to find or
  in dangerous locations!
• Other bugs probably arent going to just roll
  over and die!

12
Points from Goals

• Goals have a fixed payout
• When a bug collects the points from a goal, it
  turns red (inactive) for an amount of time
• Fixed Time
• Random Time
• Points

13
Points From Bugs

• Killing a bug by reducing its health to zero
  results in you getting half of its points
• Killing blow is the only one that counts!
• High-scoring enemies vs. low scoring enemies
• Balance and compromise

14
Staying Alive

• Health is reduced through combat
• When you get attacked health decreases by a
  random amount find out how much yourself!
• Defending can reduce the damage taken in combat
• But! Can only defend against one bug at a time!
• Counter attack fighting back
• Again, can only attack one bug at a time
• Can both defend and attack simultaneously

15
Staying Alive continued

• If victorious in combat youre still likely to be
  injured
• Health does not regenerate naturally
• Have to eat!
• Plants and dead bugs are both sources of food
• How much is healed per unit of time is again left
  to you to discover (Im still not sure!)

16
Staying Alive continued

• Plants are static
• If you saw it once, itll still be there next
  time!
• When close to a plant you can order your bug to
  eat it
• Your bug will recover some health
• The plant never dies and is never eaten up
• Plants are also always active unlike goals
  that need time to recharge before they give
  points out again

17
Staying Alive continued

• Clever strategies for combat are helpful and
  encouraged
• Sneaking up on opponents
• Lying in wait
• Camping at good spots
• Preying on the weak (mean but effective)
• Avoiding combat entirely
• And many, many more

18
Ticks And DoAction() Again

• Very simple example
• Your bug is injured and there are no other bugs
  around
• A plant is nearby and is in a quiet and safe
  position
• Probably a good idea to go and eat it!
• Call method to move towards plant
• On each subsequent turn, check if youre close
  enough to eat it
• If so, start eating it!

19
Ticks And DoAction() Again

• More complicated example
• Your bug is moderately damaged and there are no
  bugs in the immediate area
• Theres a plant nearby that is very exposed
• Another plant is further away but safer
• This Terrarium has been running for nearly the
  entire pre-set time limit and your score is quite
  low
• Heal? Score points? Fight? Run away? Spin around
  and around for the remaining time?
• Up to you!

20
Interacting With The Environment

• How do I do all these wonderful things?
• Environment is only accessible locally i.e. I can
  see (or feel) anything my bug is touching and
  anything in a small-ish arc directly in front of
  me
• However, information about what Ive seen before
  can be collected and stored for later use

21
Stored Information About The Environment

• Arguably the most important a list of all the
  goals seen to date
• Information about their location, the points they
  offer, their fixed and random times are all
  stored and can be accessed at any time
• Details of how the information is stored and how
  to access it will be discussed in subsequent
  lectures same goes for plants and bug
  information

22
Stored Information continued

• A list of all plants and their information is
  stored also
• Information about their location is generally the
  only relevant thing youll use
• Again, information can be accessed at any time
• Most likely to use it when your injured and
  looking for a place to heal

23
Stored Information continued

• Information about bugs is also stored
• Lots of information is recorded about bugs
• Health, score, kills, speed, heading, position
  etc
• When you last saw them (important!)
• Only have information about bugs youve seen
• Bugs (unlike goals and plants) are dynamic
• Probably not going to be in the same place or
  have the same information as the last time you
  saw them
• As the information gets more and more out of
  date, its likely to become less and less reliable

24
Stored Information continued

• Aside from information about the goals, plants
  and bugs within the Terrarium, another extremely
  helpful feature is the Map
• A record of the whole Terrarium, which starts
  blank but is updated as you explore
• Arranged internally in a natural way that
  corresponds to the actual layout of the Terrarium
• Can be accessed at any time and for any position
  in the Terrarium

25
More About Mapping

• Terrarium uses a positional system similar to
  Cartesian co-ordinates
• GridX and GridY
• X and Y co-ordinates like a graph
• Different origin
• All bugs, plants, goals and rocks have a position
  specified by a GridX and a GridY
• Different objects have different sizes!

26
Mapping continued

• The map is arranged identically to the Terrarium
  co-ordinate system
• The specific details of how to use and access it
  will be covered later
• Simple 2-dimensional array arrangement
• Contains information about what is in the square
• If its an organism (goal, rock etc.) it also
  stores information about the organism

27
Bug Vision Basics

• As mentioned, bugs can see and feel objects
  that are either touching them or are within a
  120 arc directly in front of them
• Arc extends for just under two bug lengths
  (roughly 8(?) GridSquares)
• Information about goals, bugs etc. is
  automatically updated every turn
• But only information about things you can
  currently see are updated!
• Beware of out-of-date information

28
Vision And Exploration

• New feature for this year
• Have I seen this square?
• Map holds a boolean (true or false) variable for
  every square in the Terrarium that is true if
  your bug has seen the square, and false if it
  hasnt seen the square
• Handy for exploration
• How to use it is again left to you

29
Moving Around a Terrarium

• Moving can occur in one of two ways
• Move to a position in a straight line
• Position can be specified exactly, or can be
  supplied by saying move to this goal, or move
  to this plant etc.
• Find a path and move along it
• Obstacles almost always disrupt movement in a
  straight line
• Automatic path-finding and following code

30
Path-Finding continued

• Based on the A Algorithm
• Specifics on path-finding and the methods used
  within Terrarium will again be covered later
• Fairly efficient method of path-finding
• Really shouldnt call it many times within a
  single DoAction()
• Fairly relaxed time-limit on thinking (but one
  exists nonetheless)

31
Path-Finding continued

• As youll see the path-following code is not
  perfect!
• Works well enough to allow you to plan and make
  journeys over long distances
• Breaks journey down into smaller, straight line
  segments automatically
• Dont worry about understanding how path-finding
  works at this stage

32
Enough On The Basics!

• That concludes the basics of Terrarium
• Almost certainly left something out though!
• Dont worry, any omissions or confusing points
  will be explained in later lectures!
• Any questions at this point?

33
Basic Strategies

• Ideas to use in Terrarium can come from a huge
  variety of places
• Articles online (and on the module page)
• Journals and books from the library
• AI techniques and approaches
• Nature what real bugs do
• Web-sites
• From nowhere! (Original ideas)

34
Strategy

• As well as short term plans (heal when Im hurt,
  attack a weak and nearby bug) you will probably
  also want to have a long term strategy
• Explore everywhere
• Kill everyone
• Hide for a long time till everyone is dead
• Etc

35
Strategy continued

• Rewarded for original, interesting and unique
  ideas
• Like Virus, it helps a lot if they actually work!
• The point of the game is to score the most points
• The point of the coursework is to be creative in
  the way you try to score points!