This talk provides an overview of AI techniques for perception and action selection in robots. Modern autonomous mobile robots, like Autonomous Guided Vehicles (AGVs), face difficulties when acting in an unstructured world, where they have to cooperate with humans. Key problems are intelligent perception and action selection. AI provides solutions to tackle these complex problems. This talk provides an overview of various AI approaches, such as behavior trees and deep (reinforcement) learning. Examples are provided from our research on autonomous mobile robots.