Artificial Intelligence Depot
Visiting guest. Why not sign in?
News, knowledge and discussion for the AI enthusiast.
FEATURES COMMUNITY KNOWLEDGE SEARCH  
Robot Building for Beginners
Not only does this book assist you in understanding component parts of robot development, but also it prepares you with techniques to learn new discoveries on your own.
More information at Amazon US UK
Beating Traffic
Prediction and Control
 
Beating Traffic

KVAL has an article on Traffic Dodger, which discusses the software. It can plan routes through towns using the vast quantities of data available, avoiding traffic jams and minimising journey time -- Hi-Tech Tool Helps Traffic Snarls. Thsi is a good example of a planning system used in practise.

TRN Magazine reports of ongoing research in Portugal about multi-agent systems and learning;
Software Agents Ask for Help. Essentially, agent select tutors, and ask for suggestions in performing actions. The task -- controlling single traffic lights -- doesn't seem ideally suited to the approach, but the idea appears to work wonders.

79 posts.
Saturday 05 October, 14:29
Reply
Autonomous Learning in MASs

The need for a common language for agent communication is mentioned. Another challenge lies in the question of how to let agents reason about their tasks. It seems that in this specific application the researchers 'hardcoded' the attribute for consideration by implementing a formula, used by an individual agent to evaluate its strategy. An interesting question here would be whether we can let an agent, either individually or in cooperation with others, try to determine what the best/most effective attributes for consideration are without 'hardcoding' this in any way. Maybe by experimenting and comparing the the system's score to a desired utility at certain intervals?

Does this sound reasonable or rather like a stupid idea?

- Bruno

4 posts.
Sunday 06 October, 14:54
Reply