Neural network training of artificial humanoid hands for deaf-blind sign language: 12th Annual IEEE Conference Mechatronics and Machine Vision in Practice, M2VIP 2005

The paper presents a concept for the design of a humanoid hand to be trained in deaf-blind sign language. A brief history of humanoid hands is discussed, followed by an overview of the first Curtin humanoid hand. Extending on several key concepts of the first Curtin hand, a new design is put forward together with key design concepts that will be essential for training the new hand deaf-blind sign language. Such design concepts include the use of Artificial Neural Networks to provide natural control of a biomechanical system to reproduce human movement, together with the use of data gloves to train the neural networks deaf-blind sign language. Finally a brief look at the other deaf-blind signing hands is presented, together with the significance of the research.