Reference no: EM133124392
Project Outline: This project should involve using the concepts of multiple agents with exhibiting structural and functional knowledge. The whole idea is to enhance the existing paper and makes improvements in areas of Uncertainty reasoning, Knowledge and belief, Meta-learning, meta reasoning, and meta-knowledge.
If the topics mentioned above are not present in the paper, we should make one attached to it, or else explain how we improved that area.
Project: Associating Structural & Functional Knowledge for Multiple Agents
The paper we have chosen: Intelligent Coordination among Multiple Traffic Intersections Using Multi-Agent Reinforcement Learning.
The points that must be checked in the project are:
• Select a particular problem in a constrained domain;
• Consider 2 agents (nodes) with specific knowledge (structural or functional);
• Assign prior incomplete knowledge on each agent relevant to the selected topic;
• The knowledge of an agent has uncertain and minimum overlapping with the other agent;
• Start a communication among the agents by using a third agent as coordinator/accumulator;
• Agents offer max info shared during the communication with the other agents;
• Locally each agent attempts to enrich its own knowledge;
• Assign probabilities of sharing knowledge and creation of new one;
• Consider p(P/Q) probabilities;
Overview:
1. Implementing an AI agent in the controllers that optimize the flow of traffic across a single intersection and then extends it to multiple intersections by considering a multi-agent setting.
2. Traffic measurements - number of vehicles and traffic density
3. Single agent utilizes a linear layer + LSTM network while a multi-agent usa a Conv-LSTM architecture for its input.
Note: For simulation used Aimsun
Attachment:- Project Outline.rar