Robots operating with intermittent connectivity must make on-the-spot decisions about what to do now and what to defer. Successful approaches combine local autonomy with connectivity-aware coordination so systems remain safe and useful when links drop. Research traditions from behavior-based robotics through probabilistic planning and resource-bounded algorithms offer tested tools for that problem.
Local decision architectures and uncertainty
Behavior-based control pioneered by Rodney Brooks at MIT emphasizes reactive layers that keep essential functions alive without remote guidance. Probabilistic methods developed by Sebastian Thrun at Stanford University, Wolfram Burgard at University of Freiburg, and Dieter Fox at University of Washington provide principled ways to represent uncertainty about state, tasks, and link quality. Integrating those ideas yields policies that rank tasks by expected value under uncertain outcomes, so a robot will favor actions that safeguard humans and equipment or preserve mission-critical resources. This prioritization reduces catastrophic failure when communication is lost and supports graceful degradation of services.
Scheduling under resource limits
Work on anytime and resource-bounded algorithms by Shlomo Zilberstein at University of Massachusetts Amherst supplies mechanisms for producing progressively better plans as compute time or connectivity permits. An anytime planner can output a quick, safe action when a link drops and refine priorities once connectivity returns. Task prioritization therefore becomes an adaptive process: immediate safety and local sensing are top priority, followed by mid-term mission objectives, and finally lower-value tasks that depend on off-board computation or coordination.
Communication-aware coordination and human factors
Decentralized protocols and opportunistic data exchange reduce reliance on continuous links. Daniela Rus at MIT and others in multi-robot systems emphasize store-and-forward, predictive scheduling, and role-assignment that exploit intermittent windows of connectivity. Human operators and affected communities should be part of the priority model: cultural expectations, territorial restrictions, and environmental stakes influence which tasks receive precedence. For example, robots deployed for disaster response in remote regions must prioritize life safety and culturally sensitive site protection over data collection tasks that can wait. Neglecting these human and territorial nuances risks technological solutions that are efficient but socially harmful.
Adopting a layered approach that blends behavior-based reflexes, probabilistic prioritization, anytime planning, and connectivity-aware communication yields robust, explainable prioritization under intermittent connectivity. This mix preserves mission value while honoring safety, cultural norms, and environmental constraints.