This paper presents an autonomous vision-based obstacle avoidance system. The system consists of three independent vision modules for obstacle detection, each of which is computationally simple and uses a different criterion for detection purposes. These criteria are based on brightness gradients, RGB (red, green, blue) color, and HSV (hue, saturation, value) color, respectively. Selection of which modules are used to command the robot proceeds exclusively from the outputs of the modules themselves. The system is implemented on a small monocular mobile robot and uses very lour resolution images. It has been tested for over 200 hours in diverse environments.

Building humanoid robots that can interact with people in the same way as people interact with people provides a new set of challenges for robotics and artificial intelligence. We report here on extending the behavior-based approach to building a human behavior level humanoid robot.

Summary form only given, as follows. In trying to build complete autonomous and self-contained mobile robots that can operate in dynamic environments the authors have developed a set of constraints on both sensing and action that are somewhat different from the traditional assumptions. The authors have built a number of small (on the order of a few pounds in weight) totally autonomous mobile robots which do real-time sensing, some including vision, as they navigate about. This paper focuses on the connection of sensing to action and how that constrains each of them.>

The author describes a carefully designed series of networks, each one being a strict augmentation of the previous one, which control a six-legged walking machine capable of walking over rough terrain and following a person passively sensed in the infrared spectrum. As the completely decentralized networks are augmented, the robot's performance and behavior repertoire demonstrably improve. The rationale for such demonstrations is that they can help identify requirements for automatically building massive networks to carry out complex sensory-motor tasks. The experiments with an actual robot ensure that an essence reality is maintained and that no critical disabling problems have been ignored. The present work is based on the drawing of analogies between evolution in the animal world and robotic evaluation.>

Most animals have significant behavioral expertise built in without having to explicitly learn it all from scratch. This expertise is a product of evolution of the organism; it can be viewed as a very long-term form of learning which provides a structured system within which individuals might learn more specialized skills or abilities. This paper suggests one possible mechanism for analagous robot evolution by describing a carefully designed series of networks, each one being a strict augmentation of the previous one, which control a six-legged walking machine capable of walking over rough terrain and following a person passively sensed in the infrared spectrum. As the completely decentralized networks are augmented, the robot's performance and behavior repertoire demonstrably improve. The rationale for such demonstrations is that they may provide a hint as to the requirements for automatically building massive networks to carry out complex sensory-motor tasks. The experiments with an actual robot ensure that an essence of reality is maintained and that no critical disabling problems have been ignored.

The MIT mobile robot project began in January of 1985 with the objective of building machines that could operate autonomously and robustly in dynamically changing environments. Five working robots, each progressively more intelligent and sophisticated, are now available. These robots are reviewed, focusing on their innovative aspects. These include novel ideas in the areas of control systems, sensors, actuators, power supplies, and power processing. Perspectives on mobile robotics are outlined, including objectives, experiences, mistakes, and future plans.>

The MIT mobile robot project has adopted a layered approach to building robust control systems for autonomous mobile robots. We build a complete control system for a very simple task achieving behavior for our robots, then lay additional complete control systems on top of it to achieve higher level tasks. We have this architecture operational on three completely different types of hardware (a general purpose computer, a special purpose multi-processor, and processor-less logic network) and are now working on a fourth, by way of a special purpose silicon compiler for it. Exactly the same high level language specification files are used on three of these four architectures. The control architect is thus freed from even the most global considerations of the target architecture.

A new architecture for controlling mobile robots is described. Layers of control system are built to let the robot operate at increasing levels of competence. Layers are made up of asynchronous modules that communicate over low-bandwidth channels. Each module is an instance of a fairly simple computational machine. Higher-level layers can subsume the roles of lower levels by suppressing their outputs. However, lower levels continue to function as higher levels are added. The result is a robust and flexible robot control system. The system has been used to control a mobile robot wandering around unconstrained laboratory areas and computer machine rooms. Eventually it is intended to control a robot that wanders the office areas of our laboratory, building maps of its surroundings using an onboard arm to perform simple tasks.

Mobile robots sense their environment and receive error laden readings. They try to move a certain distance and direction, and do so only approximately. Rather than try to engineer these problems away it may be possible, and may be necessary, to develop map making and navigation algorithms which explicitly represent these uncertainties, but still provide robust performance. The key idea is to use a relational map, which is rubbery and stretchy, rather than try to place observations in a 2-d coordinate system.

This paper describes an algorithm that decomposes the free space into nonoverlapping geometric-shaped primitives suitable for path planning. Given a set of polygonal obstacles in space, the algorithm first decomposes concave obstacles into connecting convex obstacles in order to have a uniform obstacle representation, which facilitates later processing. The neighborhood relations among these Convex obstacles are identified and then used to locate critical "channels" and "passage regions" in the free space, and to localize the influence of obstacles on free space description. The free space is then decomposed into nonoverlapping geometric-shaped primitives where channels connect passage regions. The channels are similar to the generalized cones presented in Brooks [1]. The passage regions are represented as convex polygons. Based on this mixed representation of free space, the path planning algorithm can plan path trajectories inside the channels and passage regions. The perimeter of the channels and passage regions associated with the planned path provides the boundaries of the path that the mobile robot must stay within during path execution.

A recursive cellular representation for configuration space is presented along with an algorithm for searching that space for collision-free paths. The details of the algorithm are presented for polygonal obstacles and a moving object with two translational and one rotational degrees of freedom.

ACRONYM is a comprehensive domain independent model-based system for vision and manipulation related tasks. Many of its submodules and representations have been described elsewhere. Here the derivation and use of invariants for image feature prediction is described. Predictions of image features and their relations are made from three-dimensional geometric models. Instructions are generated which teli the interpretation algorithms how to make use of image feature measurements to derive three-dimensional size, structural, and spatial constraints on the original three-dimensional models. Some preliminary examples of ACRONYM's interpretations of aerial images are shown.

Free space is represented as a union of (possibly overlapping) generalized cones. An algorithm is presented which efficiently finds good collision-free paths for convex polygonal bodies through space littered with obstacle polygons. The paths are good in the sense that the distance of closest approach to an obstacle over the path is usually far from minimal over the class of topologically equivalent collision-free paths. The algorithm is based on characterizing the volume swept by a body as it is translated and rotated as a generalized cone, and determining under what conditions one generalized cone is a subset of another.