We describe a new metaheuristic, called cooperative swarm optimization (CSO), which manages the efficient use of a group of unmanned agents performing cooperative operations. In this work we are concerned with the problem of group exploration of an uncertain environment. The CSO algorithm was developed to perform this exploration with limited inter-agent coordination and no central oversight. CSO is a swarm search algorithm that uses the mechanics of velocity to move each agent about the information space. In this way, CSO is similar to particle swarm optimization (PSO) however, CSO differs from PSO in how information is used from other agents and how the information is combined. In particular, the individual agent velocity is affected by the current agent information, the personal history of the agent, and the group's information and is weighted based on the entropy rate of change of the information space, allowing each agent to take into account the dynamic nature of the information space. New information is weighted highest, followed by changes to currently known information. Information that does not change is rated lowest. We illustrate the derivation of the algorithm and show numerical examples to demonstrate its effectiveness for use in cooperative unmanned underwater vehicle missions.