Multi-UAV exploration is often hampered by unreliable communication, restricted field-of-view sensors, and finite travel budgets. The paper introduces Dec-MARVEL, a decentralized multi-agent exploration method that operates without communication while respecting these constraints.

Dec-MARVEL is designed for teams of lightweight UAVs equipped with limited onboard cameras. By removing the need for communication, the approach avoids a common failure point in real-world deployments. It also accounts for the requirement that each robot reserve enough travel budget to return, preventing missions from ending with stranded drones.

The abstract is truncated, so details of the algorithm's implementation and experimental results are not covered here. The contribution lies in jointly addressing communication-free coordination, limited sensing, and budget-aware planning in a single framework.