Conditional Max-Sum for Asynchronous Multiagent Decision Making

TL;DR

This work addresses multiagent decision making in dynamic, open environments by extending Max-Sum to asynchronous settings through Conditional Max-Sum, implemented on Factor Graphs with a distributed broadcasting communication framework. The authors instantiate this framework in a lane-free traffic scenario, formulating dynamic lateral regions and FGs that coordinate lateral placement while a rule-based longitudinal policy maintains safety. Empirical results show that Conditional Max-Sum delivers faster, more coordinated responses and smoother lateral maneuvers, achieving higher average speeds and lower speed deviation than traditional Max-Sum, No-Max-Sum, and MOBIL baselines, especially in large-scale, open-flow settings. The approach holds promise for real-time, scalable multiagent coordination in dynamic domains and invites future exploration of external agents and alternative DCOP techniques within the same asynchronous, distributed paradigm.

Abstract

In this paper we present a novel approach for multiagent decision making in dynamic environments based on Factor Graphs and the Max-Sum algorithm, considering asynchronous variable reassignments and distributed message-passing among agents. Motivated by the challenging domain of lane-free traffic where automated vehicles can communicate and coordinate as agents, we propose a more realistic communication framework for Factor Graph formulations that satisfies the above-mentioned restrictions, along with Conditional Max-Sum: an extension of Max-Sum with a revised message-passing process that is better suited for asynchronous settings. The overall application in lane-free traffic can be viewed as a hybrid system where the Factor Graph formulation undertakes the strategic decision making of vehicles, that of desired lateral alignment in a coordinated manner; and acts on top of a rule-based method we devise that provides a structured representation of the lane-free environment for the factors, while also handling the underlying control of vehicles regarding core operations and safety. Our experimental evaluation showcases the capabilities of the proposed framework in problems with intense coordination needs when compared to a domain-specific baseline without communication, and an increased adeptness of Conditional Max-Sum with respect to the standard algorithm.

Figures (20)

• Figure 1: (a) An example of a typical FG; (b) The corresponding distributed structure for FGs.
• Figure 2: FG example of connected lane-free vehicles.
• Figure 3: Illustrative example of two agents coordinating with Conditional Max-Sum to update their decisions asynchronously.
• Figure 4: Formed lateral regions with acceleration estimates from the perspective of agent $i$.
• Figure 5: Initial placement of vehicles for the lane-free coordination problem.