I have recently begun work on programming simulations that require the individuals in the system to be aware of, and physically avoid each other while pursuing their goals. This excercise has raised several perplexing questions: -To what extent (spacially, maybe even emotionally) are individuals aware of others? -How much do individuals care about the goals of others? -How can behavioural patterns within a group be recognized? -Can these patterns be predicted? -Do groups have an "awareness" that predetermines the actions of the individuals? -If so to what extent? These are a sample of a range of problems that I find myself ill-equipped to solve. The field of cognitive science and AI seems to offer a tool set that would aid in finding answers to these and other similar questions. I am very curious to learn more about the philosophy of CS/AI in general and the particular methods of dealing with the above questions. Thank you, 3CM

"Experts" offer a bewildering range of solutions. Read up papers on "multi-agent systems" and proceedings of conferences like ICRA, IROS, ICMAS, ALIFE, AAAI, RoboCup, Autonomous Agents, ''Agents'', IJCAI...

Be prepared to see a lot of math and logic systems (including fuzzy logic) : they go down to defining Knowledge, Belief, Intention, Decision ...

[Questions (STFW) | GDNet Start Here | GDNet Search | Forum FAQ | Google | Asking Smart Questions ]
[Docs (RTFM) | MSDN | SGI''s STL | OpenGL | File formats]
[C++ Must Haves (RTFS) | MinGW | Boost | Loki | FLTK | SDL ]

Stolen from Magmai Kai Holmlor, who held it from Oluseyi, who was inspired by Kylotan...

I may have misunderstood your intent but it seems to me you want to be looking at flocking.

Your questions seem ambiguous to me.

Can you tell us in more detail what your simulation is required to do?

---

Stimulate

Thanks for responding Fruny. I have begun (and will continiue) to study the sites that you''ve directed me to. Bewildering is a bit of an understatement. The thing that has me excited is the shift from "object" to "agent" oriented thinking. Instead of objectA deciding on and doing methodX, there is agentA asking agentB what it should do. I suppose that this is the code version of two heads are better than one. It also would allow for much finer distinctions in the eventual outcome of a task. I think it also begins to answer some of my questions.

Are there any books on multi-agent systems that are worth my time and money?

Thanks again,
3CM

Fup,

My project concerns traffic analysis where the vehicles must determine: following, passing, destination awareness, parking, and extricating themselves from traffic snarls, in a high volume, and mostly unstructured roadway.

I am not too sure what you mean by flocking, but to my mind it doesn''t seem that this traffic would behave like a flock of birds with random individual movement, but group direction.

Sorry to be brief, but I''m late for rugby.

3CM

quote:

---

Original post by 3cheeseMac
Are there any books on multi-agent systems that are worth my time and money?

---

Honestly, I don''t know, I mostly use the University''s library.

The only (serious) publisher I can think of right now that probably has stuff on the topic (beside conference proceedings, which is what I''m working from) would be Springer-Verlag. They seem to have a series on agent systems.

Beware though that they are not game-oriented (if that''s what you are looking for) but more applied/research. And, well... books can be expensive (hardcover doubly so), so if you''re on an allowance, you may want to save a bit (I know I spend a lot on books).

[Questions (STFW) | GDNet Start Here | GDNet Search | Forum FAQ | Google | Asking Smart Questions ]
[Docs (RTFM) | MSDN | SGI''s STL | OpenGL | File formats]
[C++ Must Haves (RTFS) | MinGW | Boost | Loki | FLTK | SDL ]

Stolen from Magmai Kai Holmlor, who held it from Oluseyi, who was inspired by Kylotan...

There''s also another direction of research you might want to consider...

There has been some excellent work these past few years on self-organising systems...

I vaguely recall a recent article describing work done that shows that entropy CAN decrease in a system as pressure/temperature was increased... it was a very exciting result... basically there was a whole system of particles modelled in a box... and the researchers were testing a couple of very simple rules about particles trying to avoid each other... as more particles were added to the system, things got more chaotic... and then at a certain point, everything became very ordered... even static under certain conditions: grid-lock!

They used this research to explain why crowds of pedestrians self-organised into streams going in different directions and how these streams fluctuate and vary over time; as well as how grid-lock can occur. This sounds very much like what you are considering... but I could be wrong in this assumption.

My point is that you should consider self-organising systems as well as the AI areas mentioned by fruny. You might find some very good inspiration in any or all of these fields!

Cheers,

Timkin

Fruy: Is that Fruny as in "Fruny Green Eyewash"?

3CheeseMac: Yup you''re right, flocking won''t be of any interest to you! (although imagine a world in which cars flocked...)If you are looking to do this simulation for a game or within a context which only requires pseudo realism then this link may help:

http://www.gamasutra.com/features/20010124/adzima_02.htm

If it''s a proper simulation you''re after then these may prove interesting:

http://publish.uwo.ca/~jmalczew/gida_5/Pursula/Pursula.html

http://www.tfhrc.gov/advanc/agent.htm

You should find that those links give you a brief overview of current technology and provide you with plenty of references to chase.

Good luck

---

Stimulate

http://www.red3d.com/cwr/steer/
Some demos on steering & flocking. The two titled "Crowd Path Following" and "Queuing" would seem to be of direct interest to you. In addition, there are MANY links there =-)

---

"I believe; therefore, it is." -True Perception
"The Requested Information Is Unknown Or Classified" -Anonymous