This research studies trustworthy perception, distributed learning, secure communications, and intelligent collaboration for UAV swarms. It addresses node failures, identity spoofing, wireless interference, model poisoning, and inconsistent collaborative decisions in inspection, logistics, emergency response, and low-altitude monitoring missions. Federated learning, graph neural networks, multi-agent reinforcement learning, and node trust assessment are integrated to support task allocation, information sharing, formation control, and cooperative decision-making. Key topics include anomalous node detection, anti-interference communications, identity authentication, trustworthy routing, fault isolation, malicious node defense, and resilient swarm recovery.