 Computerworld is New Zealand's only specialised information systems fortnightly.
Subscribe now for $97.50 (24 issues) and save more than 37% off the cover price!
 Get the latest news from Computerworld delivered via email.
Sign up now
 Subscribe to Computerworld's
RSS newsfeed here and get news stories as they break.
Scientists around the world are using artificial intelligence software to bring them a step closer to building what they say will be the perfect robotic hand.
The artificial intelligence techniques should help researchers at the University of Portsmouth in the UK and Jiao Tong University in Shanghai in their joint effort to create software that will learn and copy human hand movements. And that should enable the robotic device to successfully mimic intricate, dexterous movements only capable today by the human hand.
Honghai Liu, senior lecturer at the University of Portsmouth's Institute of Industrial Research, told Computerworld that creating the perfect robotic hand is akin to "the Holy Grail of science."
"The human hand makes us different from animals," said Liu. "We are talking about having super-high-level control of a robotic device. Nothing that exists today even comes close. It is still a long way to go, but we are confident we are on the right track."
While the device could provide help to the manufacturing industry, Liu said it also could have significant implications for the elderly and those with disabilities.
Liu said he used what he calls a cyberglove, which is covered in tiny sensors to capture data about how the human hand moves. The cyberglove was filmed in a motion-capture suite by eight high-resolution, digital charge-coupled device (CCD) cameras with infrared illumination and measurement accuracy up to a few millimetres.
"I am interested in intelligent algorithms, which make mechanical robotic hands act with human hand capabilities," said Liu. "From an engineering and robotics perspective, a human-like robotic hand is mechanically complex [to build] because of his multiple [depths of field] and its difficult implementation. From a computing point of view, its grasp and manipulation estimation is computationally very expensive. Besides, the connection of human brain signals and hand muscles is another challenging problem we confront."
|
