We designed and implemented a distributed processing environment for uniformly communicating and sharing information among remotely located robots and human operators. This environment interconnects heterogeneous processors to withstand dynamically changing configurations and communications failures, and to support mixed-mode operation (simultaneous teleoperation and autonomous modes). The key design notion is undirected, data-driven communication: the recipients of data rather than the creators define communications paths for well-formed data blocks augmented by descriptive properties. Using this environment, we have implemented both a hierarchical path planner for an intelligent mobile robot operating over a local-area network and a telerobotic testbed that supports long-distance teleoperation.