Titlebook: Robotic Object Recognition Using Vision and Touch; Peter K. Allen Book 1987 Kluwer Academic Publishers 1987 Interpolation.Region Growing.S

hypotension

2-D Vision,cene from machine vision alone has proved to be difficult. The analogy of an image matrix to a human retina has only served to illuminate the powerful kinds of processing taking place in the visual cortex, processing that is poorly understood at present. The research of David Marr and others has tri

LARK

3-D Vision,thors the ‘2½-D’ sketch [39]. Currently, there are several sensing systems that can derive depth or surface orientation from a scene. Among these are laser range finders[1,38,64], photometric stereo [29] and binocular stereo [5]. Laser imaging is potentially hazardous and has difficulty with shiny m

窝转脊椎动物

Tactile Sensing,fer 3-D shape and structure from projected 2-D images has led most researchers to try and emulate this human information processing ability. However, the complicated interaction and coupling of surface reflectance, lighting and occlusion yield intensity arrays that machines cannot understand well. T

图画文字

Integrating Vision and Touch,cts that can be matched against the models in the model data base. The procedures described in this chapter use both sensing modalities, integrating the data from the sensors to build high level descriptions of what is seen and felt. The first integration procedure is a hierarchical procedure for bu

突变

Matching,at the object is and to determine its orientation. The matching routines try to find an object in the model data base that is consistent with the surface and feature information discovered by the sensors. The intent is to invoke a uniquely consistent model from the 3-D surface and feature primitives

卷发

Experimental Results,iments is to show that tactile sensing can reliably complement machine vision sensing and that together these two sensing modalities can create robust and accurate 3-D descriptions of surfaces and features. In particular, experiments have been run on scenes that would severely test the recognition c

interior

占卜者

Tactile Sensing,he approach taken here is that for tasks such as object recognition, vision sensing is not enough. What is needed is more sensory information that can supplement the sparse and sometimes confusing visual data. In this work, the additional sensory data is supplied by a tactile sensor that is actively controlled and is used in an exploratory manner.

avulsion

阉割

Integrating Vision and Touch, primitive used in the model data base, facilitating matching between the sensed objects and the models. The second procedure is a method for creating smoothed boundaries of hole and cavity cross sections using both vision and touch which is accurate and also useful for matching.