A process-based model of robot task plan representation is introduced, and some of its advantages are discussed. By process-based it is meant that the plan is represented as a set of concurrent communicating computational processes, rather than as a graph or a set of local propositions. A description is given of how the authors have developed process composition operators that allow them to represent explicitly and succinctly the relationship between plan components to express sensing and reaction, dynamic resource allocation, and error monitoring and recovery. The approach has formal benefits as well as representation benefits; using the techniques of the (extensive) process algebra literature, they introduce a way to analyze plans (manually or automatically) for liveness and efficiency properties and indicate how the method could help in plan generation.< >