Once More, with Feeling: Tactile Sensors for Mechatronic SystemsThursday, January 3rd, 2013
Whether robots will ever have feelings remains an open question, but a robot that can feel is already a reality.
With the help of the right types of sensors, actuators and software, robots can be given the sense of feel. Utilizing these tools, researchers at the University of Southern California’s Viterbi School of Engineering have developed a robotic finger that can quickly detect and identify a variety of natural materials by their textures. Taking a cue from nature, the researchers modeled their BioTac robotic finger to mimic a human finger.
The device features a soft, flexible skin that’s wrapped over a liquid filling. Besides possessing the ability to identify a wider array of materials than a human finger, the device can also detect where and in which direction forces are being applied to the fingertip, as well as the temperature of the object being touched.
Fingerprint Detects Vibrations
BioTac shares an important characteristic with human fingers: a fingerprint, which is used enhance the device’s sensitivity to vibration. As the robotic finger glides across a textured surface, its skin vibrates in specific ways. These vibrations are detected by a hydrophone inside the bone-like core of the finger. A human finger relies on similar vibrations to identify textures, but the robotic finger is even more sensitive.
BioTac’s sensory modalities are made possible by three separate sets of transducers. The device’s skin, a low-cost, molded elastomeric sleeve, can be easily replaced if it suffers some form of physical damage or simply wears out. The robotic finger’s sensors, electronic circuitry and connections are all protected inside a rigid core.
Inspired By 18th Century Theorem
Developed by Gerald Loeb, a USC professor of biomedical engineering, and recently graduated doctoral student Jeremy Fishel, BioTac is designed to translate unique vibrations into identifiable code that identifies specific materials with the help of a specially-designed algorithm. The algorithm was inspired by a theorem developed by Thomas Bayes, an 18th century mathematician, that describes how decisions can be made from the exploratory information humans obtain while attempting to identify an object by touch.
During development, BioTac was trained on more than 100 ordinary materials collected by the researchers from nearby fabric, stationery and hardware stores. By the time the training was completed, the robotic finger was able to correctly identify randomly selected materials 95 percent of the time, requiring an average of five exploratory movements on each sample. In tests using pairs of materials featuring similar textures, BioTac only rarely failed to correctly identify a material and then only when human subjects, making their own exploratory movements, could not distinguish a difference at all.
Loeb and Fishel predict that BioTac will pave the way for new applications in a variety of fields, including intelligent prostheses, personal assistive robots and consumer product testing. The co-inventors are now partners in SynTouch, a Los Angeles-based company that develops and manufactures tactile sensors for mechatronic systems that mimic the human hand.
Founded in 2008 by researchers from USC’s Medical Device Development Facility, the start-up is currently selling BioTac systems to other robotics researchers as well as to manufacturers of industrial robots and prosthetic hands. A BioTac Evaluation Kit, also now available, contains a BioTac finger and the accessories and other materials required to acquire sensory data on a Windows PC. The company also offers software and programming libraries that are designed to help customers develop custom applications based on BioTac.