Reasoning robots. The art and science of programming robotic agents (Q2487869)

This book presents the `fluent calculus' as one possible approach to solve the frame problem in artificial intelligence. The formal model for reasoning about actions is extended into actual programming languages and systems for the design of robotic agents. The PROLOG-based method FLUX is one example for these systems and supports the problem-driven top-down design of robotic agents with the cognitive capabilities of reasoning, planning and intelligent troubleshooting. The book starts with a careful introduction into the fluent calculus, based on an axomatic predicate-logical formalism and the presentation of the logic programming system FLUX. Based on this knowledge the reader can find many interesting problems from AI and how they are dealt with, such as incomplete states and incomplete knowledge (which makes the approach more realistic). Planning, nondeterminism, imprecision, indirect effects (ramification), the acquisition problem (troubleshooting) and the inclusion of all these topics into robotics fill the next chapters. Each chapter follows a very strict construction: it starts with an axiomatic logical model of the problem area, supported by illustrative examples which are mostly given by predicate calculus formulas and program code. Exercises and biographical notes conclude the respective representation. The book is very understandable and recommendable for anybody who wants to solve AI problems by means of predicate logics and logic programming. It shows that this approach is more and more applicable to the solution of real-world problems. It will be an excellent textbook for appropriate AI or robotics courses and very appropriate for graduate students.