| 78 | |
| 79 | ---- |
| 80 | [http://us.blizzard.com/en-us/games/sc/ StarCraft] |
| 81 | |
| 82 | [http://skat.dnsalias.net/mburo/sc2011/ StarCraft AI Competition] |
| 83 | |
| 84 | [http://overmind.cs.berkeley.edu/ Winning team] |
| 85 | |
| 86 | '''Basic strategy'''[[BR]] |
| 87 | |
| 88 | '''Production:''' |
| 89 | - first strategy was to hard code plan as finite state machines. |
| 90 | e.g. Build tanks unless the enemy builds air units, then build anti-air units. If the enemy builds cloaked units, build detectors, and so on. |
| 91 | this strategy did not work because if the opponent prevents the agent from gathering the necessary resources with gas steal, the agent will be stuck. |
| 92 | - second strategy: instead of providing a fixed sequence of units and structures to produce, provide a combination of units and structures. |
| 93 | The planner will choose the best unit or structure to build at any moment given its resources. |
| 94 | If opponent tries gas steal, the agent will expand base or build units that don't require gas until the problem is resolved. |
| 95 | - scouting: scout enemy base to see what units are being produced and plan accordingly. |
| 96 | |
| 97 | '''Controlling units''': |
| 98 | - hard to build control nodes for units, therefore there is a high cost for building more than one type of unit. |
| 99 | focused on mutalisks. large numbers are hard for humans to control without clumping. easy prey for area attacks. not a problem for computer. |
| 100 | potential fields balancing attractive forces on targets and repulsive forces on threats move the mutalisk around |
| 101 | use map editor to build stage for agent to run through different combat scenarios and learn optimal potential field strengths |
| 102 | - predict results of actions based on damage calculations. assign values to targets and mutalisks based on resource costs. |
| 103 | uses this information to choose which targets to attack. |
| 104 | |
| 105 | '''Path-finding:''' |
| 106 | - keeps updated map of of last known location of enemy units. combines this map with know movement and attack speed of units to create threat map. |
| 107 | threat map is combined with path-finding algorithm. areas of high threat for a unit will have a high cost to traverse. |
| 108 | |
| 109 | ---- |
| 110 | |
| 111 | |
| 112 | == RPG == |
| 113 | |
| 114 | [http://nwn.bioware.com/ Neverwinter Nights] |
| 115 | |
| 116 | - RPG base on Dungeons & Dragons |
| 117 | - Requires purchase of game |
| 118 | - Can be modified for many purposes |
| 119 | http://edugamesresearch.com/blog/category/neverwinter-nights/page/2/ |
| 120 | |
| 121 | [http://nwn.bioware.com/downloads/standaloneserver.html#download Standalone server][[BR]] |
| 122 | Not sure how useful this is. Downloaded and ran. Got this |
| 123 | |
| 124 | Neverwinter Nights Server |
| 125 | Build:8109 |
| 126 | Copyright BioWare Corp 1998-2004 |
| 127 | |
| 128 | Server: Loading...[[BR]] |
| 129 | Server: Running... |
| 130 | |
| 131 | Server: Unable to use port 5122 |
| 132 | |
| 133 | Work on AI in NPCs[[BR]] |
| 134 | [http://www.bth.se/tek/jhg.nsf/bilagor/bachelor_thesis_revised_bertil_jeppsson_pdf/$file/bachelor_thesis_revised_bertil_jeppsson.pdf AI-controlled life in Role-playing games][[BR]] |
| 135 | [http://members.home.nl/rcpcollaris/documenten/ScriptieV5.4.pdf An Adaptive Automatic Pilot for Role Playing Games 5.4][[BR]] |
| 136 | |
| 137 | ---- |
| 138 | |
| 139 | == FPS == |
| 140 | |
| 141 | [http://www.udk.com/ Unreal] |
| 142 | |
| 143 | - Windows only |
| 144 | - See Walter |
| 145 | |
| 146 | ---- |
| 147 | |
| 148 | == Life Simulation == |
| 149 | |
| 150 | [http://www.thesims3.com/ The Sims] |
| 151 | |
| 152 | - Requires purchase. |
| 153 | - NPC have pretty sophisticated AI |
| 154 | |
| 155 | Not sure how this game would work. |
| 156 | |
| 157 | ---- |