A Multi-Agent System (MAS) is a computational environment, such as the internet or smartgrids, where individual software agents interact with each other, sometime in a cooperative manner, sometimes in a competitive manner, and sometimes autonomously pursuing their individual goals, accessing resources and services of the environment and working towards the goals of the entities that these agents represent. Application areas include transportation, logistics, computer games, films, ...
In a Multi-Agent System, agents have no direct control over one another. As such, they must persuade their peers to act in certain ways. One of such persuasion techniques is negotiation. Negotiation enables groups of agents to arrive at a mutual agreement regarding some belief, goal or plan. The negotiation process can take many different forms, such as auctions, protocols in the style of contract net, and argumentation.
Negotiation in Multi-Agents Systems is one of IIIA’s most prolific lines of research, having studied and developed different techniques and negotiation systems, including HANA and NB3.
HANA: Human-Aware Negotiation Architecture. HANA is an agent architecture suitable for multiple bilateral negotiations in realistic problems involving humans. The architecture deals with pre-negotiation and provides a new search and negotiation technique where search and negotiation go hand in hand: the former providing offers to propose, and the later providing commitments for pruning the search space, and information for fine-tuning the evaluation of offers. The architecture represents graded beliefs, dynamic desires and general intentions. It can be extended incorporating new behavioural models that can enrich the negotiation strategy with new information.
NB3: Negotiation-Based Branch and Bound. NB3 is a new family of negotiation algorithms that is applicable to domains with many agents, an intractably large space of possible agreements, non-linear utility functions and limited time so an exhaustive search for the best proposals is not feasible. It assumes that agents are selfish and cannot be blindly trusted, so the algorithm does not rely on any mediator. This family of algorithms applies heuristic Branch & Bound search to find good proposals. It uses a negotiation protocol that imposes almost no restrictions and is therefore also applicable to negotiations with humans.