Robotic search requires deciding where to look next as evidence arrives, while every motion command spends travel distance and mission time. This thesis studies that decision problem under a shared Bayesian active-search formulation and develops learned policies for real-time single-robot field execution and simulated coordination across heterogeneous teams.

The first contribution is a path-aware single-robot policy. We introduce a path-integral expert that scores complete shortest-path routes by their accumulated expected information gain, then amortize that expensive expert into a Graph Attention Network policy by behavior cloning. The resulting policy achieves up to a 254x speedup over the path-integral expert and an 11x speedup over a one-step greedy expert, transfers without retraining across the tested map geometries and object densities, and yields 1.10x to at least 2.86x faster mission completion against a strong NATS baseline in field tests on an autonomous ground vehicle in a 75,000-square-meter forested environment.

The second contribution extends the same shared-belief interface from single-robot search to coordinated teams. We propose Q-learning for Uncertainty-Guided Efficient Search Teams (QUEST), a multi-agent Q-learning framework that trains a related graph-policy family but optimizes a per-decision Bellman target over downstream belief, robot positions, team coverage, and remaining budget. The formulation covers single robots, homogeneous UGV teams, and heterogeneous UAV–UGV teams.

In a pooled evaluation of 1,000 episodes, QUEST reaches a 0.910 F-score with four UGVs versus 0.875 for the strongest learned baseline—a lead that grows with team size. Furthermore, the policy maintains its lead under communication outages and two simultaneous mid-mission robot failures without retraining, and in out-of-distribution UAV–UGV compositions, it improves the early-budget F-score (0.759 vs. 0.728) while using 10% less path length and 12% less duplicate coverage. A map-structure analysis shows that gains on the four-map suite concentrate on constrained, branching maps under binding budgets, while a broader map sweep establishes that algebraic connectivity and modularity are descriptive lenses rather than predictors of performance.

All team results are simulation studies; the field evidence in this thesis is the single-robot deployment. Together, these policies connect real-world field deployment with coordinated team search that generalizes across composition, communication, and partial failure under one shared-belief active-search formulation.