Trends in robotics

Robotics in manufacturing. " The phrase invokes visions of large groups of manipulators performing welding on an automobile assembly line. We also visualize the replacement of a worker by a robot to create islands of automation, a generally unsuccessful approach to using technology in manufacturing, which may have contributed to the collapse of the U.S. robotics industry in the 1980s. We associate this picture with traditional manufacturing engineering and not computer science. However, there is another side to contemporary industrial robotics. Many aspects of robotics, particularly those involving computer vision, part representation for grasping, or robot programming are properly the province of computer science, and play an important role in factory automation. Furthermore, as robotic systems become more intelligent and perform more complex cognitive tasks, the computer science aspects will become more and more crucial. There is a major revival underway in factory robotics. Robot sales in the U.S. have increased by about 30% per year for the past two years and show evidence of an even greater gain this year. There are many reasons for this resurgence. As U.S. industry downsizes and reduces its workforce to regain global competitiveness, automation in all forms becomes more important. Industrial robots have become more reliable, and their cost has decreased in real dollars during the past decade. But there are other factors: Today's robots are more intelligent ; they are equipped with better vision systems and other sensors; their programming languages are easier to use (so that off-line robot programming is now commonplace); and they can be interfaced more easily with CAD/CAM systems. All these factors involve applications of computer science. There is also a major emphasis on flexible, small-lot manufacturing in contemporary industry , which requires rapid changeover to new products. Such flexibility often depends on the use of computer-controlled robots. One of the fastest growing applications of robotics is in the area of assembly. Nearly all PC board assembly is done by robots, and they are moving into assembly applications in many industries. Computer science viewpoints are affecting such aspects of the assembly field as sequence planning, part orientation, and part feeding. I foresee future 3D solid modeling systems being tightly coupled with AI-based planners that will yield efficient (perhaps "optimum" in some sense) assembly sequences for the products being represented. Further, the assembly planner will generate assembly instructions, suitable for either humans or robots, as appropriate. Among the important …