Managing Automation- July 2008

is they are very inflexible, and track and workflow process changes can cost hundreds of thousands of dollars. About a year ago, however, a new company entered the scene. Seegrid, a company with an idea that originated in the labs of Carnegie Mellon University, created an autonomous robot that can interpret its surroundings and move on its own. The technology includes a vision system that takes pictures of the environment and software that interprets and models the surroundings. The robot first learns about its environment from a human, who leads it down the specific path it needs to go, and instructs it to perform certain actions at points along the path, such as stop and pick up a pallet. The information is stored in memory — basically a computer hard drive — which enables the robot to move down every aisle or path it has been instructed to travel. The current versions of the Seegrid robots, which include the model GP8, a motorized pallet jack, and the GT3, which can pull carts (replacing a forklift, for example), can travel 15 miles and store more than 10,000 paths in memory. In the future, as computing power rises, these robots will be able to travel 30 miles and understand hundreds of thousands of paths, the company says. “Our goal is to create an autonomous robot that can think, see, and move on its own,” says Greg Cronin, Seegrid’s executive vice president. Of course, the robot is not thinking, Cronin concedes; it just reacts to paths defined by humans. But as probabilistic algorithms increase with computing power, robots will start to perceive the environment better and interact more with humans and with other systems. “Most robots don’t interact with the environment because in classical industrial applications of the 1980s, robots were nothing more than pre-programmed automated machines used to execute repetitive tasks without any adaptability,” Siciliano says. However, using cameras, range finders, sophisticated software, and sensors — or even artificial skin — for tactile perception, robots will soon become more human-like. “The modern robot is a customizable robot which the worker can use as a tool,” Siciliano says. For example, a robot could be taught to handle a pair of scissors for cutting fabric. While some early work is under way, a robot’s ability to manipulate objects requires a high degree of sensory fusion coupled with artificial intelligence, which is still a bit futuristic. For now, flexible robots are used in ware- houses, but even that is a big step forward, as they are finally cost-effective. “Robots have been used for a long time to assemble cars. There are a lot of things that make a robot cost-effective for building a $35,000 car, but not cost-effective for moving pallets of laundry detergent,” Genco’s Rector says. As the technology has evolved, and become affordable, it has put robots within reach of companies that are finding new applications. “There is certainly a place where robotics can perform tasks that don’t need to be done by people, such as the repetitive moving of [products] from point A to point B that doesn’t take a lot of skill,” Rector says. “Robots can do that and leave the things that require skill to humans.” Another new AGV vendor is tackling the warehouse with swarms of robots. Five-yearold start-up Kiva Systems has created a massively parallel processing material handling system that uses robots working collectively to Seegrid’s autonomous robots can travel 15 miles and store more than 10,000 paths in memory. July 2008 27

Table of Contents Feed for the Digital Edition of Managing Automation- July 2008

Managing Automation- July 2008 Contents Take 1 HP's Acquisition of EDS May Face Integration, Global Delivery Challenges Rimini Plans to Woo SAP Support Users Rockwell Steps Up Its Activities in Software Microsoft Targets the Process of Innovation Supply Chain Standouts Honored by AMR Notes Who Are Today's Influential Thinkers? The Robot Revolution Is the Price Right? Without a Trace Harvesting the Suggestion Box Product Scan Advertiser Index Next

For optimal viewing of this digital publication, please enable JavaScript and then refresh the page. If you would like to try to load the digital publication without using Flash Player detection, please click here.