If you are looking for an answer to the question What is Artificial Intelligence? and you only have a minute, then here's the definition the Association for the Advancement of Artificial Intelligence offers on its home page: "the scientific understanding of the mechanisms underlying thought and intelligent behavior and their embodiment in machines."
However, if you are fortunate enough to have more than a minute, then please get ready to embark upon an exciting journey exploring AI (but beware, it could last a lifetime) …
In the design and construction of mobile robots vision has always been one of the most potentially useful sensory systems. In practice however, it has also become the most difficult to successfully implement. At the MIT Mobile Robotics (Mobot) Lab have designed a small, light, cheap, and low power Mobot Vision System that can be used to guide a mobile robot in a constrained environment. It is our belief that the constraints of many environments will allow the implementation of a system that provides vision based guidance to a small mobile rover. The purpose of our vision system is to process realtime visual input and provide as output information about the relative location of safe and unsafe areas for the robot to go.
This missive is being written for a workshop on autonomous vacuum cleaners. The problem statement reads as follows: ..this symposium will concentrate on AI as applied to a physically instantiated robot for vacuuming household floors. The target problem is to autonomously vacuum a living room while doing the right thing with furniture, trash, pets, etc. I will argue informally that this is not an AI problem at all. None of the techniques currently being investigated by researchers can produce effective vacuuming behavior better than the simple expedient of robust mechanical design.
Each definition requires different capabilities in the vacuuming agent and hence different internal software architectures. However, the definitions suggested form a progression from reactive, to synthetic, to intelligent, and the lessons learned looking at one problem supply important insights for tackling the next. The paper then describes the Animate Agent project which defines an architecture for robot control that attempts to embody both the reactive methods of control required in simple vacuum cleaners and the symbolic methods of situation classification and plan choice required by synthetic vacuum cleaners.
What reasoning mechanism will a roomvacuuming robot need? One might argue that no logic at all is needed, and this would be hard to refute. The matter is too open-ended for hard and fast proofs. But until someone designs a highly successful vacuuming robot that eschews all logic, it is not unreasonable to consider what sort of logic such a robot might use. Indeed, even if an alogical robot could be perfected, this is no argument that another design might not make good use of logic.
INTRODUCTION Many seemingly complex behaviors observed in animals are actually simple reactive behaviors to sensory stimulus. It is believed by many researchers that all motion in animals is activated by combinations of these low-level behaviors. Moreover, community activities, such as gathering in groups or cooperative searching for food has also been successfully demonstrated as using simple primitive behaviors. Many robotic tasks can be accomplished using the same types of simple reactive primitives. However, the environment to which animals have adapted does not always map easily to "structural" types of robot tasks.
Several hundred such operating system functions and utilities are embedded into the system to simplify third party applications development and system portability for alternate embedded applications. This will permit the very rapid development of new high level applications by university researchers and OEM's using the basic building blocks, subsystems, software, sensors and development tools already developed by Cyberworks. Long Term Test Results The above architecture has been implemented on three diverse types of platforms for real world evaluation. The first is a fully autonomous vacuum cleaning robot. "CyberVac" is a robotic vacuum cleaner with self-navigational capabilities so advanced that it can be placed in a complex room and navigate around furniture in such a way that the entire floor surface of the room is methodically and efficiently cleaned without preprogramming and without an inefficient "first pass" to "learn" the room.
Designing the ultimate autonomous vacuum cleaner is the quantifiable and clearly-defined goal of this symposium. It falls short of addressing the problems which most interest roboticists working in artificial intelligence. The painstaking progress and restricted performance of current mobile robots apparently necessitates such a low-level review. Overcoming such challenges and setbacks is essential if robots are to become capable of operating in more interesting regimes. Several questions are presented to help evaluate whether a robotics research program remains oriented towards meaningful long-term goals of autonomy, intelligence and capable performance.
Autonomous vacuuming agents are no exception. Under the assumptions that the physical architectures for such agents include an on-board, self contained power supply, (to avoid the cord-entanglement problem), and sensors which may be prone to failure and noise, the basic problem is one of tracking and resource management. An accurate and timely representation of critical parameters and system state must be maintained if the task is to be accomplished with any efficiency or reliability. Bayesian techniques provide the most effective and accurate means to maintain this information. Most of these techniques were developed in the early 1960's.