Universal Robotics
From Wikipedia, the free encyclopedia
| Universal Robotics, Inc. | |
|---|---|
 |
|
| Type | |
| Founded | 2001 |
| Headquarters | Nashville, Tennessee  |
| Key people | David Peters, CEO Richard Alan Peters II,CTO |
| Industry | Automation Software |
| Products | Neocortex |
| Website | http://www.universalrobotics.com |
Universal Robotics is a software engineering start up, taking to the materials handling market a revolutionary self-learning machine intelligence operating system. Called Neocortex, Software with an IQ, it is not an AI type of technology, (though it offers a solution where AI has not worked). It is based on the learning behavior of mammals and is a new algorithmic paradigm called Natural Intelligence. This invention is patent protected and was created at Vanderbilt University. It has been the "brain" of NASA's humanoid robot, Robonaut, for three years.[1]
Contents |
[edit] Product
Neocortex is an operating system software for mobile machines. It will automate equipment which has heretofore required human operators. The real-time self-learning qualities of the product allow machines to successfully function in dynamic, sophisticated environments. The software becomes more intelligent with experience, meaning its utility will expand accordingly. This represents a new application paradigm. The first two applications of Neocortex will be automating mix boxed palletization and depalletization using six-axis industrial robots and automating, by degree, discreet forklift operations, building up over time to a fully automated forklift.
[edit] Background
Machine intelligence as applied to industry, has been hampered by years of false starts and dead ends. Till now, efforts have failed to create [machines]] which could make novel choices in dynamic environments. The problem stemmed from the belief that if enough facts where loaded into a sophisticated data base, a machine would become intelligent (i.e. the Artificial Intelligence approach).[2] Since this approach was not working, a number of roboticists in academia came to believe that a machine could exhibit intelligent behavior only through physically manipulating the world and, through its own sensors, learn the immediate effects of its actions. Intelligence would emerge as the machine developed skills through its interactions with people and the world.[3]
Beginning his research over seven years ago, Dr. Richard Alan Peters and his team designed algorithms for an architecture which show the theory is correct. Neocortex does not rely on classical artificial intelligence (AI), and though it organizes sensory-motor data in vector space, it is not simply an artificial neural network. It is modeled after hypotheses on the acquisition of natural intelligence by animals, integrating concepts from a number of disciplines concerned with behavior, biology, and engineering.[4]
[edit] Operations
Universal is a hybrid of Functional and Product organizational structures. Functionally, there six task areas: 1) strategic planning, 2) sales and service of customers, 3) engineering and programming, 4) quality control, 5) research and development, and 6) security. The sophistication of Neocortex, the significant state-of-the-art engineering, and the small staff, support a Product structure approach, where by layers of management are reduced and the interconnectedness of individual tasks is paramount.
[edit] References
- ^ Ambrose, Rob, et al. (Jul 2000). "Robonaut: NASA's space humanoid" 15 (4): 57-63.
- ^ Davidson, Keay. "Why 'AI' won't come true any time soon", San Franciso Chronicle, 2001-6-20.
- ^ Brooks, Rodney (1990). "Elephants Don't Play Chess". Robotics and Autonomous Systems 6: 3-15.
- ^ Kawamura, Kazuhiko, et al. (2000). "ISAC: Foundations in human-humanoid interaction". IEEE Intelligent Systems 15 (4): 38-45.
