Execution-Aware Dynamic-Priority Scheduling for Functionally Heterogeneous Multi-Robot Disaster Response

Multi-robot disaster-response task allocation requires balancing functional matching, task value, and execution efficiency. This study proposes an execution-aware heterogeneous dynamic-priority scheduling framework (HDPS) that connects interpretable task scoring with navigation feedback. Tasks are represented by location, type, value, remaining urgency, task state, and current owner. Robot-specific scoring weights are updated from candidate-pool statistics through an exponential moving average, while completion, repeated navigation failure, and stalled progress trigger task-state updates or reassignment. A local encounter-coordination layer supports execution. The evaluation considers task-level functional-role heterogeneity on physically identical robots, with pre-generated task sets rather than continuous task arrivals or evolving disasters. Nine methods are compared on ten paired ROS 2-Gazebo-Nav2 scenarios, giving 90 main runs. Every method completes all 15 tasks in each main run. Relative to the nearest and no_type methods, HDPS improves the specialization match rate by 50.00 and 50.67 percentage points, respectively, with positive paired differences in all ten scenarios. Separate ablations identify the value term as the main contributor to early high-value service: the top-five-task completion rank improves from 8.32 without value scoring to 5.62 with HDPS. HDPS retains an effective value completion rate of 88.0% while achieving a mean total completion time of 178.87 s, compared with 88.9% and 193.85 s for fixed weights; the time difference is descriptive rather than statistically significant. These results support a practical trade-off among functional matching, early service of high-value tasks, and overall task-clearing time within the evaluated setting.