Nova
NovaAuthor(s):
Alberto Uriarte
Affiliation:
Drexel University
Country:
USA Race:
Terran
BWAPI version:
3.7.4
Type:
DLL
Download:
Language:
C++
Source code:
Links
Nova is a bot based on a multi-agent system. It plays StarCraft: Brood War and was developed by Alberto Uriarte and uses the BWAPI framework to talk to the StarCraft engine.
Nova uses a multi-agent system that splits the work into micro-management tasks, which control a large army with different abilities (tactical decisions), and macro-management tasks, which take decisions about economy, building construction and technology research (strategy decisions). At the moment Nova only plays the Terran race.
Features
- Blackboard architecture
- Working memory
- Terrain analysis (BWTA2 project)
- Opponent build order prediction
- Opponent's military units tracking (threat map)
- Flocking squad behaviours
- Score system for target selection
- Influence maps
- Finite state machine
- Build map information
- Graphical user interface to debug
- Monte Carlo Tree Search for squad movement (it uses a combat simulator)
Videos
Downloads
- Source code, under the GNU GPL v3: bitbucket.org/auriarte/nova, or a copy of the old version here: Nova-src.zip (22 MB)
- Binaries: Nova-exe.zip (old version), or the competition DLL in the infobox
Related publications
- (2016) Improving Monte Carlo Tree Search Policies in StarCraft via Probabilistic Models Learned from Replay Data
- (2015) Automatic Learning of Combat Models for RTS Games
- (2014) Game-Tree Search over High-Level Game States in RTS Games
- (2014) Walling in Strategy Games via Constraint Optimization
- (2013) A Survey of Real-Time Strategy Game AI Research and Competition in StarCraft
- (2012) Kiting in RTS Games Using Influence Maps
- (2011) Multi-Reactive Planning for Real-Time Strategy Games
The full list is on the author's page.
Competitions
| Year | Place | Event | Result |
|---|---|---|---|
| 2014 | 13th | AIIDE 2014 | 32.40% |
| 2014 | 8th | CIG 2014 | 38.89% |
| 2013 | 7th | AIIDE 2013 | 33.81% |
| 2013 | 7th | CIG 2013 | 32.1% |
| 2012 | 8th | AIIDE 2012 | 41.61% |
| 2012 | 5th | CIG 2012 | 53.1% |
| 2011 | 8th | AIIDE 2011 | 47.50% |
| 2011 | Eliminated in the first round (bracket A) | CIG 2011 | |
| Medals: none | |||
Results by opponent race
| Tournament | vs Zerg | vs Protoss | vs Terran | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Win | Loss | Win rate | Win | Loss | Win rate | Win | Loss | Win rate | |
| 2011 AIIDE competition | 38 | 52 | 42% | 51 | 129 | 28% | 82 | 8 | 91% |
| CIG 2011 | 8 | 2 | 80% | 0 | 30 | 0% | 0 | 0 | - |
| 2012 AIIDE competition | 184 | 0 | 100% | 361 | 927 | 28% | 144 | 40 | 78% |
| CIG 2012 | 79 | 11 | 88% | 228 | 312 | 42% | 115 | 65 | 64% |
| Student StarCraft AI Tournament 2012 | 4 | 4 | 50% | 12 | 6 | 66% | 17 | 5 | 77% |
Press
- (9/26/11) Inteligencias artificiales en forma de bots compitiendo en StarCraft (in Spanish)
- (9/26/11) "Programar bots de Starcraft puede llegar a ser más complejo que hacerlo para el ajedrez", entrevista a Alberto Uriarte (in Spanish)