1 CO3301 - Games Development 2 Week 16 Blackboard Model Gareth Bellaby.

Slides:



Advertisements
Similar presentations
The Enterprise Skills Story
Advertisements

Design, prototyping and construction
Heuristic Search techniques
1 CO Games Development 2 Week 15 Terrain Analysis Gareth Bellaby.
RECAP! SKILL and NATURAL ABILITIES –Speed reaction time –Agility –Co-ordination –Flexibility & Balance Differences in novice and skilled performers:- consistency,
MILITARY SYMBOLOGY NEXT. Today you will receive a class on military symbology. At the end of this class you will be able to properly identify five basic.
Using Cellular Automata and Influence Maps in Games
International Plekhanov Conference Research of Successful Communications Planning; Team Restructuring.
1 SYS366 Week 1 - Lecture 2 How Businesses Work. 2 Today How Businesses Work What is a System Types of Systems The Role of the Systems Analyst The Programmer/Analyst.
WEEK 4 Material Lecture 4a (Wed.). Use Cases/Actors o What is a use case ? l A sequence of actions performed by a system that yields an observable result.
CMSC 414 Computer and Network Security Lecture 9 Jonathan Katz.
Week 8 Implementation Design Alex Baker. Implementation Design System Design – Describes what the system should do Implementation Design – Describes what.
Neural Networks (NN) Ahmad Rawashdieh Sa’ad Haddad.
Adapted from J. Scott Armstrong for use in lectures related to Persuasive Advertising. File: Target-market Research-R10 Lectures for AdPrin/Testing Ads.
Regression testing Tor Stållhane. What is regression testing – 1 Regression testing is testing done to check that a system update does not re- introduce.
Leadership & Team Building
Copyright © 2008 Allyn & Bacon Meetings: Forums for Problem Solving 11 CHAPTER Chapter Objectives This Multimedia product and its contents are protected.
©Ian Sommerville 2004Software Engineering, 7th edition. Chapter 15 Slide 1 Real-time Systems 1.
thinking hats Six of Prepared by Eman A. Al Abdullah ©
Jon Curwin and Roger Slater, QUANTITATIVE METHODS: A SHORT COURSE ISBN © Cengage Chapter 3: Using Graphs.
Decision Making Dr Vasuprada Kartic NAC Batch IX PGDCPM.
Chapter 14: Artificial Intelligence Invitation to Computer Science, C++ Version, Third Edition.
Lecturer: Gareth Jones Class 8: Persuasive Messages.
Comp 20 - Training & Instructional Design Unit 6 - Assessment This material was developed by Columbia University, funded by the Department of Health and.
1 CO Games Development 2 Week 19 Probability Trees + Decision Trees (Learning Trees) Gareth Bellaby.
1 CO Games Development 2 Week 17 – Goal-Oriented Action Planning Gareth Bellaby.
Problem Solving Design Loop What are the steps? How many are there? What is the purpose of this???
EMOTIONAL RESPONSE Section B – Spectatorship Topics
PERSUASION.
CE Operating Systems Lecture 3 Overview of OS functions and structure.
7 Systems Analysis and Design in a Changing World, Fifth Edition.
Decision Support Systems (DSS) Information Systems and Management.
CO1301: Games Concepts Dr Nick Mitchell (Room CM 226) Material originally prepared by Gareth Bellaby.
Games Development Game Architecture: Entities CO2301 Games Development 1 Week 22.
The characteristics of skilful movement. Key processes in physical education Developing skills in physical activity. Making and applying decisions. Developing.
GAME PLAYING 1. There were two reasons that games appeared to be a good domain in which to explore machine intelligence: 1.They provide a structured task.
An Introduction to Formative Assessment as a useful support for teaching and learning.
Fall 2015 ECEn 490 Lecture #8 1 Effective Presentations How to communicate effectively with your audience.
A Roadmap towards Machine Intelligence
Dr Nick Mitchell (Room CM 224)
Unit – I Presentation. Unit – 1 (Introduction to Software Project management) Definition:-  Software project management is the art and science of planning.
1 CO Games Development 1 Week 3 Game Agents 2 Gareth Bellaby.
Bloom’s Taxonomy Dr. Middlebrooks. Bloom’s Taxonomy.
LHCbComputing Computing for the LHCb Upgrade. 2 LHCb Upgrade: goal and timescale m LHCb upgrade will be operational after LS2 (~2020) m Increase significantly.
Planning and Organizing Chapter 13. The Planning Function Planning for a business should stem from the company’s Business Plan – The business plan sets.
ENG101B Week 8 EXECUTIVE SUMMARY (adapted from “Writing Executive Summaries” written by James C. Morrison, Lecturer in the Department of Urban Studies.
 You are to create an original response to the play you have been studying in class.  A creative response is an original and unique piece of creative.
Presented by The Solutions Group Decision Making Tools.
1 CO Games Development 2 Week 13 Influence Maps Gareth Bellaby.
Loop Design What goes into coding a loop. Considerations for Loop Design ● There are basically two kinds of loops: ● Those that form some accumulated.
Lecturer: Gareth Jones Class 12: Persuasive Messages.
CO Games Development 2 Week 16 Blackboard Model
Unit 4 Working With Communities
Lecture 3: Procedures of extension work
CREATE - ASSESSMENT LEVEL GRADE CREATE: Your final Game design.
CO Games Development 2 Week 22 Trees
Unit 5 Working With Communities
Real-time Software Design
Design, prototyping and construction
CO Games Development 2 Week 19 Extensions to Finite State Machines
CIS 488/588 Bruce R. Maxim UM-Dearborn
Regression testing Tor Stållhane.
Foundations for making smart decisions
CO Games Development 1 Week 8 Depth-first search, Combinatorial Explosion, Heuristics, Hill-Climbing Gareth Bellaby.
Plan your journey.
Games Development Game Architecture: Entities
Implementation of Learning Systems
Design, prototyping and construction
Building progressions for accuracy and tactical awareness
Presentation transcript:

1 CO Games Development 2 Week 16 Blackboard Model Gareth Bellaby

2 Blackboard The blackboard model is a decision making method. It is a way to coordinate different agents, routines or experts.

3 Blackboard Based on the simple metaphor of a blackboard with people around the blackboard discussing a problem. The problem, and all workings out, are written on the blackboard. The insight is that a collective understanding of a problem may be better than an individual understanding. It may be more efficient to have many experts each with a partial understanding of a problem than one expert who has a full understanding.

4 Architecture

5 Example For example, a unit needs to move from point A to point B, through enemy fire, and with a river in the way. No specialist understands the whole problem. Specialist 1 understands movement under fire. Specialist 2 understandings river crossings. The problem is written on the blackboard. Specialist 1 signals its relevance to movement under fire. The unit then reaches the river. Specialist 2 signals its relevance to river crossings. The unit crosses the river. Specialist 1 signals it relevance to movement under fire.

6 Specialists The specialists are components that can operate on the data written up on the blackboard. The area of expertise of each specialist is narrow. A specialist may indicate a relevance value, indicating how relevant it thinks its expertise is to a problem. No communication is allowed between the experts. Everything goes through the medium of the blackboard: the common message area. The specialists can work with partial solutions.

7 Arbiter The arbiter selects which of the specialists to execute, i.e. it decides which specialist to accept the advice of. The arbiter might just choose the specialist with the highest relevance. A more sophisticated arbiter might take short-term and long-term relevance into account, or use motivation models such as emotional state, e.g. how excited they are, see the Pandemonium example given at the end of the lecture.

8 Architecture Two alternatives: 1) Multiple specialists each with their own area of expertise. It is assumed that only one specialist at a time will be dealing with a specific problem. 2) Specialists with overlapping areas of expertise. More than one specialist could signal relevance. More then one specialist could be allowed to deal with a problem at a time.

9 Example A problem is posted onto the blackboard. Specialist 1 has no understanding of the problem currently on the blackboard but does understand another type of problem. Specialist 2 has a partial solution to the existing problem so writes up its partial solution. Specialist 1 now recognises the new problem and can provide a final solution. This is an example of alternative 1. The movement under fire with a river-crossing example is also a version of this approach.

10 Characteristics The blackboard model offers flexibility. The order of reasoning is not pre-determined. At any given point during the process the most relevant specialist is selected. This allows forward-chaining (from data to goal) and backward-chaining (from goal to data) all together and in any sequence. Specialists can act in a variety of ways, e.g. request more data, perform an action, act as input to another system (even another blackboard).

11 Characteristics A specialist need not know how its assertions or signals are going to be used. A specialist does not care who will use its output, e.g. it could go to another specialist, or it could be the final output of the system. A specialist need not know the origin of an assertion or signal. The specialist is only concerned with fulfilling a request.

12 Extensions to the approach Can order the information. Can identify information as data (input to the system) or as goals (assertions generated within the system). Can ascribe a credibility score to assertions.

13 Multiple Agent Example This is based on Isla and Blumberg. RTS game. Missions which requires skills, e.g. plan assault, provide supporting fire, assault attack, reconnaissance, defend area from aircraft. Units which have several skills, e.g. a planning skill, combat skill, build skill, etc.

14

15 RTS Example Slide 8 describes two alternative blackboard approaches. The RTS scenario is an example of alternative 2: specialists with overlapping areas of expertise. More than one specialist could signal relevance. More then one specialist could be allowed to deal with a problem at a time. For example, a sapper (combat engineer) unit has the attack skill and the build skill. An infantry unit has the attack skill and this would be rated higher (it is more proficient) than the sapper. The infantry unit may have a basic build skill - they could entrench, put up sandbags, etc - but they would be less proficient than the sapper at building.

16 RTS Example Note the way that scheduling is dealt with. The blackboard lists the current missions. Imagine that there are two build missions on the blackboard and that two units are available: a sapper and an infantry unit. The sapper would take the build mission with the higher priority. However, the infantry unit would take the second mission since it has the build skill even though it's not particularly good at building.

17 Combining Data The blackboard approach includes the idea choosing between competing alternatives. Can use thresholds to trigger actions, e.g. a particular threshold is needed before the action is triggered. Can combine several stimuli. The simplest way to do this is to sum the stimuli together. For example, imagine identifying the approach of a tank. If the tank is in clear view then this is a strong visual stimulus which would trigger our response. However, imagine that the tank is partially obscured. If we had a some other partial stimulus, such as sound, then the two in combination cause the our response to be triggered.

18 Pandemonium A different metaphor used to describe a variant upon the blackboard approach. The blackboard is set up as normal. There are lots of little demons who read the black as per the basic metaphor. However, the demons become excited if they see something they understand. Imagine them jumping up and down, and jumping more if they view something they understand. The most excited demon is chosen. We can remember the other demons as well so if the apparent best choice fails, we can select the next demon in sequence.

19 Further reading Rabin, S., (2003), "Common Game AI Techniques", AI Game Programming Wisdom 2, Charles River Media: Hingham, Mass, USA. Isla, D., and Blumberg, B., (2002), "Blackboard Architectures", AI Game Programming Wisdom, Charles River Media: Hingham, Mass, USA.