This article proposes local bidding strategies for autonomous agents participating in an auction-based multi-agent coordination system, in order to improve the scalability and reactivity of the architecture in large-scale coordination scenarios. Based on a careful analysis of the reactivity requirements and the computational costs of the central auctioneer (costs for solving Linear Integer Programs) and the local agents (costs for path-planning and task execution), this article explores the idea of each agent bidding for a subset of available tasks that are locally relevant to the agent. Each agent first employs a computationally light euclidean distance-based and percentile-based screening method to choose a subset of available tasks, followed by a more computationally complex, but realistic path-planning based cost-estimation and bidding for the chosen subset. The proposed strategy not only reduces the overall computational cost at the agents, but also at the central auctioneer, by reducing the size of the combinatorial optimization problems and the overall communication requirements of the architecture, thereby improving the scalability and reactivity of the overall system. It is shown that, through a one-time simulation-guided design of the bidding parameters, the improved reactivity and scaling is achieved while retaining the optimality or near-optimality of the resulting task-allocation. The performance of the proposed bidding strategies is evaluated in two large-scale simulation scenarios and the reduction in computational costs and the near-optimality of the task allocation is demonstrated.

In this article, we present a reactive multi-agent coordination architecture for the management of material flows between production/pickup stages and delivery/drop-off stages, in scenarios such as underground mines and automated factory floors. The pickup and delivery stages are modelled as variable task queues, with no a priori information about the inflow into the production queues. The proposed solution coordinates the movement of a group of mobile agents operating between the two stages in a reactive and scalable manner, so that the material is transported from multiple production queues to multiple delivery queues in a balanced/equalized manner. In such a scenario, centralized planners suffer from low reactivity and poor scaling, as the number of agents and number of queues increases. To overcome this problem, we propose a decentralized approach comprising of two separate auction-based task distribution systems for the production and delivery stages, along with behavior-tree based management of agent autonomy and task bidding. Each auction system tracks the length of production/delivery queues and solves the optimal task assignment, based on the bids submitted by the agents. The agents participate in one of the two auction systems at any given time, based on the status of the behavior tree executing the two-stage tasks. We analytically show that the proposed decentralized auctioning approach along with agent autonomy and bidding managed by behavior trees, offers better scalability and reactiveness compared to the centralized approach. The proposed methodology is experimentally validated in a lab environment, in three illustrative material flow management scenarios, using TurtleBot3 robots as agents.