Robot Vision

Citation

Alboul, L. (2011), "Robot Vision", Industrial Robot, Vol. 38 No. 6. https://doi.org/10.1108/ir.2011.04938faa.007

Robot Vision

Robot Vision

Article Type: Book review From: Industrial Robot: An International Journal, Volume 38, Issue 6

Edited by Ales Ude,InTech,2010,Free download,614 pp.,ISBN: 978-953-307-077-3,Web site: www.intechopen.com/books/show/title/robot-vision

The book is dedicated to Machine Vision with a focus on real-life robotic applications, and covers various aspects of this rapidly expanding field.

Both industrial and mobile robots are explored, including teleoperated robots, autonomous unmanned ground vehicles (UGV), unmanned aerial vehicles (UAV), and wheeled and biped walking robots.

The book consists of 30 chapters presenting a wide range of robot vision technologies. These include systems for industrial inspection, manufacturing guided by vision, vision-based mobile robot navigation, object tracking and identification, object detection in vision search, scene description, grasping using vision, and biomedical applications. Both hardware and software aspects of robot vision are presented.

One drawback is that the chapters are not organised into separate sections.

More than one-third of the articles in the book deal with vision-based mobile robot navigation from various perspectives. Navigation with the use of omnidirectional vision and navigation involving only a single camera or multi-camera set ups are described, to name a few. Among new methods are vision based navigation inspired by visual sonar-based reactive navigation, and navigation steering strategies for UAVs by using only a single perspective camera.

Several papers are dedicated to vision motion analysis for safe 3D navigation in unknown and dynamic environments. Various methods are described, such as motion analysis based on optical flow and removal of unwanted motion through image stabilisation in active vision.

Machine intelligence is also tackled; and, based on the assumption that intelligence is associated with sophisticated memory, a method of using sparse distributed memory (SDM) to navigate a robot is proposed. SDM is a kind of associated memory proposed by Kanerva (Kanerva, P., Sparse Distributed Memory, MIT Press, Cambridge, MA, 1988).

Vision based systems for acquiring 3D information, including stereo-vision in real time, structure from motion and time of flight cameras, are discussed as well.

Several papers present new methods to work with advanced vision systems, such as dioptric and hyperbolic catadioptric systems, in order to facilitate object detection, to ensure an integrated perception, or to focus on an optimal design of a system to minimise position errors.

A foveated vision system, similar to human eyes, is presented and shown to be beneficial for object recognition.

One article is dedicated to the development of new materials to achieve the required optical properties of lenses.

Systems where the data obtained from vision sensors are fused with the non-visual data are also discussed.

Specialised applications, such as pose estimation through 2D-3D registration for orthopaedic surgery, are discussed as well. Biometric applications are also tackled.

One chapter is concerned with the interpretation of human motion, such as recognition of gait types. Another chapter describes a guideline for testing virtual reality systems in robot tele-operation.

Industrial applications include robotized manufacturing, visual servoing and an active security system for a robot manipulator.

The variety of techniques that are used are well established, such as support vector machines (SVM) for recognition, principal components analysis (PCA), Kalman filters, back- propagation, wavelet transforms; but, in general, they are extended or modified to satisfy the system requirements.

The book can be interesting for specialists in robot vision and for those who want to keep themselves updated on the latest research in the field.

Lyuba AlboulSheffield Hallam University, Sheffield, UK