Robust multi-agent decision making in faulty environment

A brief overview is given of current state of the art in robust multiagent decision making in a faulty environment (distributed consensus); and also a family of novel algorithms is presented to achieve consensus that are deterministic, simple, require single-bit messages, and can be implemented in hardware. For such systems to work properly, the issues of reaching common decision (consensus) in the presence of faults have to be addressed. The authors offer a structural solution to this problem. The approach is illustrated on a hypothetical example of a number of autonomous robots in manufacturing that all have to agree on a common decision determined by the values of some central controllers.