An intelligent agent must be capable of using its past experience to understand how its actions affect the world. Given some objective, the agent must use its understanding to produce a plan that is robust to uncertainty. This book presents a novel computational framework called the Adaptive Modelling and Planning System (AMPS) that attempts to meet these requirements for intelligence. AMPS generalises from limited experience by grouping together similar states and actions using a process called abstraction. Several different abstraction approaches have been proposed in past, but they generally only increase resolution, require a large amount of data before changing the abstraction, do not generalise over actions, and are computationally expensive. AMPS aims to address these issues by continuously adapting the abstraction in response to experience as it is accumulated and replanning in the regions of the state space where it is needed most.