This selective vision is based on an understanding and analysis of the driving task. We illustrate the effectiveness of request-driven routines by comparing the computational cost of general scene analysis with that of selective vision in simulated driving situations."

Intelligent robotics has become the focus of extensive research activity. This effort has been motivated by the wide variety of applications that can benefit from the developments. These applications often involve mobile robots, multiple robots working and interacting in the same work area, and operations in hazardous environments like nuclear power plants. Applications in the consumer and service sectors are also attracting interest. These applications have highlighted the importance of performance, safety, reliability, and fault tolerance. This volume is a selection of papers from a NATO Advanced Study Institute held in July 1989 with a focus on active perception and robot vision. The papers deal with such issues as motion understanding, 3-D data analysis, error minimization, object and environment modeling, object detection and recognition, parallel and real-time vision, and data fusion. The paradigm underlying the papers is that robotic systems require repeated and hierarchical application of the perception-planning-action cycle. The primary focus of the papers is the perception part of the cycle. Issues related to complete implementations are also discussed.

Active vision and perception for resource-constrained autonomous vehicles, such as small ground robots and quadrotors, are limited in their allowable algorithmic complexity and slow reaction times. For an autonomous mobile robot to safely and reliably perform a useful task or behavior, real-time visual perception that informs a controller with a fast reaction time is needed. This dissertation covers new research developments in the areas of active vision, planning, and control for directional sensors with a focus on event-cameras and RGB cameras. Event-cameras, also known as neuromorphic cameras, are biologically inspired visual sensors that measure local changes in light intensity, mitigating latency and redundant data. Several high-level active vision algorithms, interfaced with autonomous vehicle controllers, are developed for event-cameras and quantitatively compared to analogous RGB camera algorithms, in terms of both accuracy and computational cost. In particular, motion-based perception algorithms for object recognition and tracking, action recognition, and depth estimation are developed for use on a moving quadrotor tasked with reacting to the perceived environment. Novel active vision algorithms for RGB cameras are also developed in which an autonomous ground vehicle or quadrotor interact with a human target of interest using novel action recognition and tracking perception capabilities paralleled with new control methods for target following. Furthermore, a novel occlusion-avoiding path planning algorithm that is applicable to both event-cameras and RGB cameras is developed. The proposed method computes a closed-form collection of subsets of the sensor's configuration space, referred to as visibility regions, that quantify the visibility of targets subject to the sensor field of view geometry and line of sigh visibility. This method is quantitatively compared to several existing sensor path planning methods in terms of analytical computational complexity, experimental path performance, and experimental computational cost analysis. The results of this work enable active vision, perception, and planning for resource-constrained mobile robots equipped with directional sensors such as an event-camera or RGB camera.

One of the series in Machine Perception and Artificial Intelligence, this book covers subjects including the Harvard binocular head; heads, eyes, and head-eye systems; a binocular robot head with torsional eye movements; and escape and dodging behaviours for reactive control.

Visual perception is one of the most important sources of information for both humans and robots. A particular challenge is the acquisition and interpretation of complex unstructured scenes. This work contributes to active vision for humanoid robots. A semantic model of the scene is created, which is extended by successively changing the robot's view in order to explore interaction possibilities of the scene.

This unique book explores the important issues in studying for active visual perception. The book’s eleven chapters draw on recent important work in robot vision over ten years, particularly in the use of new concepts. Implementation examples are provided with theoretical methods for testing in a real robot system. With these optimal sensor planning strategies, this book will give the robot vision system the adaptability needed in many practical applications.