She said Intel researchers from the National University of Singapore, the bus on the membership of the Association of Intel computing research neural “INRC”, revealed the promising results on the development of sensors for the sense of touch and vision and integrate them with the processor computing a nerve that evolved in the robots.
Research focuses on the role of the sense of touch in improving the capabilities of robots and their performance compared with existing systems based on the sense of vision is just, next over the superiority of processors, the computing Neuron on the structure of the governance processes of traditional hand processing of sensory data.
Commenting on this research, said Mike Davis, director of the laboratory of Intel computing neural: “the proposed research of the National University of Singapore overview mission about the future of robotics, where sensor information processing as based on events and combines multiple ways. Add this research to a wide range of research findings which demonstrate that he can assess for nerve achieve significant gains both in terms of latency and power consumption when you re-design the entire system in accordance with the model based on events includes the sensor, data types, algorithms, and engineering hardware.”
For his part, said Harold Soh, assistant professor in the Department of Computer Science, Faculty of computing of the National University of Singapore: “we are very pleased with the results that indicate that computing system nerve is one of the most promising technologies in the integration of multiple sensors with the aim of improving cognitive abilities for robots. This research is an important step towards building robots reliable, high efficiency in energy consumption, which can respond rapidly and appropriately in the operating conditions is not expected.”
The importance of research scientists, Singapore and the impact of Intel
Think robots in their work on visual processing of vision and image analysis, but with the New Discovery will move the domain to a different. Where the sense of touch and its ability high to distinguish between surfaces that differ from one layer only one of the molecules and thus give a greater ability to distinguish surfaces and materials of texture, not of form only.
Fly researchers at the National University of Singapore to change that by artificial skin that was developed recently, which contend that it helps to sensor the feel of things at over 1,000 times the sensory system human nervous, beside its ability to determine the shape and hardness of objects at a speed exceeding 10 times that of the twinkling of an eye.
Product improvement in the performance of the robots
Intel says about this in her statement: “the development of something like a sense of touch of humans when robots can contribute to improving their current performance to a large degree, has lead to the emergence of new use cases for robots. For example, can premium Android the provider of wholesale artificial easily adapt to changes in behavior that are produced in the source using the sensor touch to turn things unfamiliar and constipation by the appropriate degree of pressure to prevent slipping. May contribute to provide the robots the ability to perceive their surroundings and feel better in enhancing the level of interaction with the human and make them safer, especially in applications requiring the delivery of care, or the automation of surgical procedures, providing surgical robots a sense of touch that is missing currently.”
The cost of the company: “the development of artificial skin a big step towards achieving this vision, but it requires the provision of a slide processing can use the accurate results based on the data of the sensors of the touch in actual time, as well as high efficiency of energy consumption, allowing the possibility of fitted directly inside the robot.”
In the Benjamin T., assistant professor in the Department of materials science and engineering, National University of Singapore and National University of Singapore Innovation, Health and technology: “contribute to the provision of artificial skin high-speed processing of about half of the problems that prevent make robots more intelligent. You also need robots to artificial blood gives her the ability to cognition and learning, which are essential to achieve this goal. Our research unique about the possibility of merging the meeting system based on artificial intelligence with the processors computing the neural – like processor Louie ends (Loihi) from Intel – a major step towards enhancing the efficiency of energy consumption has robots expand their use.”
A glimpse of the sea as the company published in its statement:
The team began the researchers of the National University of Singapore to explore the possibilities of computing technology of the nervous system in processing sensory data extracted from the sensors using Touch Bar Louie ends (Loihi) research from Intel, with the aim of opening new horizons in the field of strengthening the capacity of cognitive robots. During their initial experience, the researchers used the robotic hand with wholesale artificial reading in Braille, data transfer the touch to the processor Louie are through cloud computing to improve the outcrops minute I felt a hand to the meaningful words. Success Wizard Luo are the classification of wars of Braille accurately up to 92%, along with reducing the energy consumption at a rate at least 20 times about the processor von Neumann (Von Neumann).
Based on this experience, a team of researchers of the university to improving the cognitive abilities of the robot by integrating data visual and tactile within a network nervous skeptical. To achieve this goal, the cost of robot researchers to the classification of fluid present within the vessels of non-transparent by using the input sensors of the artificial skin, and a special camera to monitor the events. The researchers used the same sensors for touch and vision to test the ability of the perception system to determine the movement of the slip ring, which represent an essential element to prevent the robot the ability to hold things continuously.
Once they collect sensory data, the panel sent to the processor graphics core and slice the Luo are research from Intel to compare the processing capabilities between them. These results, presented at the conference Robotics: Science and systems which takes place during the current week, and that the integration of vision-based events with the data of tactile sensor by using neural network so excited to have contributed to enhancing the accuracy of classification of objects by 10% compared with systems based on vision only. Moreover, the researchers noted the potential of the most promising computing technologies of the nervous system in support of robotic devices, where the managed partner Louis ends the processing of sensory data at a speed exceeding by 21% compared with the best graphics processors, with reduced energy consumption at a rate of less than 45 times.