ABSTRACT

Surfaces ........................................................................................................... 241 19.5.4 Curvature Computation and Corner Detection .................................... 243 19.5.5 Matching Polygon Fragments .............................................................. 246 19.6 Matching Models and Scene Description ........................................................ 247 19.6.1 Hypothesis Generation ........................................................................ 247 19.6.2 Mutual Compatibility and Multiple Matches ...................................... 247 19.7 Matching Procedure for Compatible Match .................................................... 248 19.8 Experimental Results and Discussion ............................................................. 252 19.9 Conclusion ....................................................................................................... 252 Keywords ................................................................................................................. 257

19.1 INTRODUCTION

The task of robots in automatic assembly and inspection requires object selection and determination of its orientation in an intelligent manner. Usually assembly machines

expect uniform parts in precise locations and use elaborate jigs. But correct placement of objects and its ideal viewing under noisy and occluded situation present difficulties and decrease system flexibility. Multiple objects under occlusion cause significant problems in identifying and locating them in the workspace of the robot. Further the enormous amount of data makes the exhaustive search for template matching impractical. To reduce the data using features like area, perimeter, length, and breadth can be applied. Position and orientation of the object do not affect these features. A decision tree can determine the objects present by indicating which measurements to make based on the values of the other measurements. The measurements quickly narrow down the possible objects that might appear in the scene.