I am a doctoral researcher working on cooperative autonomy for heterogeneous teams of Autonomous Underwater Vehicles (AUVs), focusing on multi-agent task allocation under weak underwater acoustic communication. Because subsea links are bandwidth-limited, high-latency, lossy, and intermittently connected, I study how to make coordination reliable without assuming perfect connectivity.
My work centers on distributed task allocation, especially CBBA and its improved variants, for dynamic missions with online task insertion and constraints such as time windows, coupled tasks, and multi-phase workflows (e.g., search-and-confirm). I design communication-aware mechanisms—topology-sensitive rules, local conflict resolution, and partial reallocation—and benchmark them against optimization baselines (MILP/CVRP).
I build high-fidelity simulation pipelines that integrate realistic AUV motion constraints with configurable communication models (range, delay, packet loss, bandwidth), supported by 3D trajectory and network-topology visualization. I also explore CTDE multi-agent RL (e.g., QMIX) with explicit message passing and task-conditioned attention for selective information fusion under partial observability.
I am a doctoral researcher working on cooperative autonomy for heterogeneous teams of Autonomous Underwater Vehicles (AUVs), focusing on multi-agent task allocation under weak underwater acoustic communication. Because subsea links are bandwidth-limited, high-latency, lossy, and intermittently connected, I study how to make coordination reliable without assuming perfect connectivity.
My work centers on distributed task allocation, especially CBBA and its improved variants, for dynamic missions with online task insertion and constraints such as time windows, coupled tasks, and multi-phase workflows (e.g., search-and-confirm). I design communication-aware mechanisms—topology-sensitive rules, local conflict resolution, and partial reallocation—and benchmark them against optimization baselines (MILP/CVRP).
I build high-fidelity simulation pipelines that integrate realistic AUV motion constraints with configurable communication models (range, delay, packet loss, bandwidth), supported by 3D trajectory and network-topology visualization. I also explore CTDE multi-agent RL (e.g., QMIX) with explicit message passing and task-conditioned attention for selective information fusion under partial observability.