Multi-robot task-allocation through vacancy chains

Existing task allocation algorithms generally do not consider the effects of task interaction, such as interference, but instead assume that tasks are independent. That assumption is often violated in multi-agent systems, e.g., in the case of cooperative mobile robots, where interaction effects can have a critical impact on performance. Modeling the effects of the interactions within a multi-agent system, the group dynamics, is difficult due to their complexity. The same complexity also makes it difficult to program, by hand, optimal solutions to multi-robot task allocation (MRTA) problems. We formalize the concept of group dynamics in the traditional framework of scheduling and show that task allocation in multi-agent systems with significant performance effects from the group dynamics is an NP-complete problem. We then present a simplified model of task allocation in multi-agent systems based on vacancy chains. A vacancy chain is a resource distribution process commonly found ...

• Type: Conference
• Year: 2003
• Where: ICRA
• Authors: Torbjørn S. Dahl, Maja J. Mataric, Gaurav S. Sukhatme
• Keywords: ICRA 2003, Multi-agent Systems, Robotics, Task Allocation, Task Allocation Algorithms