We describe our experiences in teaching introductory AI and in writing a textbook for the course. The book tries to make the concepts of AI more concrete via two strategies: relating them to the student's existing knowledge, and using examples based on an agent operating in an environment. Perceived Problems with Current AI TextsIn the dozen or so times we have taught introductory AI, we have used several of the existing texts, and have always had complaints from students. In a recent student evaluation survey at Berkeley, the text for the AI course was ranked lowest of all texts in computer science courses. Other instructors we have talked to share this sentiment. Some say the current texts are too shallow, or that they present too many ideas without enough motivating examples. The chapters often come across as separate, unrelated subjects, and students don't know what technique to apply to a new problem. Outsiders have criticized AI for concentrating on toy domains, and insiders complain that the texts perpetuate this perception by devoting so much space to Eliza, GPS, and other toy programs of the 1960s. Finally, the texts often have gratuitous differences in terminology and notation that make AI appear alien to the well-rounded computer science student.In reaction to these problems, we have written a text 1 that we believe presents the field in a much better light. The text has now been used in over a hundred courses and has been very well received. In this paper we outline the key pedagogical ideas behind it.2 Unified presentation of the field Some texts are organized from a historical perspective, describing each of the major problems and solutions that have been uncovered in 40 years of AI research. Although there is value to this perspective, the result is to give the impression of a dozen or so barely related subfields, each with its own techniques and problems. We have chosen to reinterpret some past research and show how it fits within a common framework and how it relates to other work that was historically separate. Often this inwflves regularizing the notation to emphasize similarities rather than differences. In some cases, we omit work that was important in its day but has since been superseded. 1Artificial Intelligence: A Modern Approach, Prentice Hall. 3 Emphasis on Relations to Previous ExperienceIn all cases, we emphasize the connections that AI has to other areas of computer science, and, to a lesser extent, to mathematics, linguistics, psychology, and philosophy.For example, many of our students have had a compiler course, and know all about BNF, LR(k) grammars, and various parsing algorithms. A course that teaches ATN grammars fails to connect to these students' previous knowledge. We present parsing by first using BNF notation, and then showing what additions are needed to handle semantics, pragmatics and disambiguation. We recognize that ATNs have an important historical place in AI, but we treat them in a Bibliographical and Historical Notes section rather than in the main text. On the o...

Abstract-The need for combined task and motion planning in robotics is well understood. Solutions to this problem have typically relied on special purpose, integrated implementations of task planning and motion planning algorithms. We propose a new approach that uses off-the-shelf task planners and motion planners and makes no assumptions about their implementation. Doing so enables our approach to directly build on, and benefit from, the vast literature and latest advances in task planning and motion planning. It uses a novel representational abstraction and requires only that failures in computing a motion plan for a high-level action be identifiable and expressible in the form of logical predicates at the task level. We evaluate the approach and illustrate its robustness through a number of experiments using a state-of-the-art robotics simulator and a PR2 robot. These experiments show the system accomplishing a diverse set of challenging tasks such as taking advantage of a tray when laying out a table for dinner and picking objects from cluttered environments where other objects need to be re-arranged before the target object can be reached.

Articles WINTER 2015 105A rtificial intelligence (AI) research has explored a variety of problems and approaches since its inception, but for the last 20 years or so has been focused on the problems surrounding the construction of intelligent agentssystems that perceive and act in some environment. In this context, the criterion for intelligence is related to statistical and economic notions of rationality -colloquially, the ability to make good decisions, plans, or inferences. The adoption of probabilistic representations and statistical learning methods has led to a large degree of integration and crossfertilization between AI, machine learning, statistics, control theory, neuroscience, and other fields. The establishment of shared theoretical frameworks, combined with the availability of data and processing power, has yielded remarkable successes in various component tasks such as speech recognition, image classification, autonomous vehicles, machine translation, legged locomotion, and question-answering systems. Copyright AI MAGAZINEAs capabilities in these areas and others cross the threshold from laboratory research to economically valuable technologies, a virtuous cycle takes hold whereby even small improvements in performance have significant economic value, prompting greater investments in research. There is now a broad consensus that AI research is progressing steadily, and that its impact on society is likely to increase. The potential benefits are huge, since everything that civilization has to offer is a product of human intelligence; we cannot predict what we might achieve when this intelligence is magnified by the tools AI may provide, but the eradication of disease and poverty is not unfathomable. Because of the great potential of AI, it is valuable to investigate how to reap its benefits while avoiding potential pitfalls.Progress in AI research makes it timely to focus research not only on making AI more capable, but also on maximizing the societal benefit of AI. Such considerations motivated the AAAI 2008-09 Presidential Panel on Long-Term AI Futures (Horvitz and Selman 2009) and other projects and community efforts on AI's future impacts. These constitute a significant expansion of the field of AI itself, which up to now has focused largely on techniques that are neutral with respect to purpose. The present document can be viewed as a natural continuation of these efforts, focusing on identifying research directions that can help maximize the societal benefit of AI. This research is by necessity interdisciplinary, because it involves both society and AI. It ranges from economics, law, and philosophy to computer security, formal methods, and, of course, various branches of AI itself. The focus is on delivering AI that is beneficial to society and robust in the sense that the benefits are guaranteed: our AI systems must do what we want them to do.This article was drafted with input from the attendees of the 2015 conference The Future of AI: Opportunities and Challenges (see Acknowledgements),...

This talk proposes a very simple "baseline architecture" for a learning agent that can handle stochastic, partially observable environments. The architecture uses reinforcement learning together with a method for representing temporal processes as graphical models. I will discuss methods for learning the parameters and structure of such representations from sensory inputs, and for computing posterior probabilities. Some open problems remain before we can try out the complete agent; more arise when we consider scaling up. A second theme of the talk will be whether reinforcement learning can provide a good model of animal and human learning. To answer this question, we must do inverse reinforcement learning: given the observed behaviour, what reward signal, if any, is being optimized? This seems to be a very interesting problem for the COLT, UAI, and ML communities, and has been addressed in econometrics under the heading of structural estimation of Markov decision processes. Learning in uncertain environmentsAI is about the construction of intelligent agents, i.e., systems that perceive and act effectively (according to some performance measure) in an environment. I have argued elsewhere Russell and Norvig (1995) that most AI research has focused on environments that are static, deterministic, discrete, and fully observable. What is to be done when, as in the real world, the environment is dynamic, stochastic, continuous, and partially observable? This paper draws on a variety of research efforts supported by NSF (IRI-9634215), ONR (N00014-97-1-0941), and ARO (DAAH04-96-1-0341).In recent years, reinforcement learning (also called neurodynamic programming) has made rapid progress as an approach for building agents automatically (Sutton, 1988;Kaelbling et al., 1996;Bertsekas & Tsitsiklis, 1996). The basic idea is that the performance measure is made available to the agent in the form of a reward function specifying the reward for each state that the agent passes through. The performance measure is then the sum of the rewards obtained. For example, when a bumble bee forages, the reward function at each time step might be some combination of the distance flown (weighted negatively) and the nectar ingested.Reinforcement learning (RL) methods are essentially online algorithmd for solving Markov decision processes (MDPs). An MDP is defined by the reward function and a model, that is, the state transition probabilities conditioned on each possible action. RL algorithms can be model-based, where the agent learns a model, or model-free-e.g., Q-learning citeWatkins:1989, which learns just a function Q(s, a) specifying the long-term value of taking action a in state s and acting optimally thereafter.Despite their successes, RL methods have been restricted largely to fully observable MDPs, in which the sensory input at each state is sufficient to identify the state. Obviously, in the real world, we must often deal with partially observable MDPs (POMDPs). Astrom (1965) proved that optimal decisions in POMDPs depend on ...

No abstract