Comparative Study of Market-Based and Threshold-Based Task Allocation

Abstract

In this paper we compare the costs and benefits of market-based and thresholdbased approaches to task allocation in real world conditions, where information and communication may be limited or inaccurate. We have performed extensive comparative experiments in an event-handling domain. Our results indicate that when information is accurate, market-based approaches are more efficient; when it is not, threshold-based approaches offer the same quality of allocation at a fraction of the expense. Additionally, both approaches are robust to low communication and task perception ranges in our experimental domain.

Bio-Inspired Communication for Self-Regulated Multi-Robot Sytems

“…Here we have attempted to scrutinize how MRTA solutions have been affected by the variations in communication strategies. Kalra & Martinoli (2007) empirically studied the comparative performance of MRTA under both predefined and self-organized approaches with event-driven simulations. They found that the accuracy of information is crucial for predefined market-based approach where every robot communicated with every other robot.…”

The effect of load on agent-based algorithms for distributed task allocation

“…Note that our hybrid algorithm shares similarities with the threshold-based R-WASP algorithm studied by Cicirello and Smith [6,7], Nouyan et al [24], and to the market-based algorithm with a "reserve price" studied by Kalra and Martinoli [17]. All three algorithms have the ability to select mail (similar to threshold-based algorithms) and to assign mail to the individual who is best suited to it (similar to the market-based algorithms).…”

“…In particular, it is important to consider under what types of conditions an algorithm loses performance, and what features of the algorithm cause this. Kalra and Martinoli [17] take this approach, investigating the effect of communication range and state estimation on threshold-and market-based algorithms. They find that market-based algorithms perform better under perfect state estimation and good communication conditions, but break down in poor conditions.…”

Multi-Level Partitioning and Distribution of the Assignment Problem for Large-Scale Multi-Robot Task Allocation

“…When range limitations of physical sensors mean that data for utility estimation is only collected with acceptable accuracy for "nearby" tasks (cf. [4]), then the sparsity of the utility matrix follows from the relative positions of robots and tasks. As demonstrated by the results in Section VI-A, spatial calculations mean that nearby information often contributes most significantly to the final task allocation solution; this is a useful property because estimates of utilities for far away tasks are also likely to be poorer than those nearby.…”

“…Decentralized/Distributed approaches do not distribute the utility information globally, instead individual agents may have little or no dependence on other robots. Some algebraic, greedy, market-based, and swarm-intelligence algorithms fall into this category (see e.g., [3,4,5]). Important prior work has looked at parallel implementations of known centralized algorithms (e.g.,, [6,7]) and under a variety of differing communication constraints (e.g.,, [8]).…”