Cooperation of Multiple Autonomous Robots and Analysis of their Swarm Behavior

TL;DR

This paper introduces a novel partition-based modeling approach for analyzing cooperative autonomous robot swarms, enabling verification of complex interactions and irregularities under environmental and swarm size changes.

Contribution

A new partitioning method for modeling robot interactions that reduces complexity and allows formal verification of swarm behavior and correctness.

Findings

Modeling advanced robot interactions with limited state combinations

Detection of irregularities like deadlocks and non-termination

Verification techniques using model checking for swarm systems

Abstract

In this paper, we extended previous studies of cooperating autonomous robots to include situations when environmental changes and changes in the number of robots in the swarm can affect the efficiency to execute tasks assigned to the swarm of robots. We have presented a novel approach based on partition of the robot behavior. The sub-diagrams describing sub-routs allowed us to model advanced interactions between autonomous robots using limited number of state combinations avoiding combinatorial explosion of reachability. We identified the systems for which we can ensure the correctness of robots interactions. New techniques were presented to verify and analyze combined robots' behavior. The partitioned diagrams allowed us to model advanced interactions between autonomous robots and detect irregularities such as deadlocks, lack of termination etc. The techniques were presented to verify and analyze combined robots' behavior using model checking approach. The described system, Dedan verifier, is still under development. In the near future, timed and probabilistic verification are planned.