Congrats to U Penn and Berkeley for their Best Paper Award!Monday, May 13th, 2013
Researchers at University of Pennsylvania’s Haptics Group and UC Berkeley’s ICSI have announced amazing progress in teaching a robot to learn haptic adjectives of real-world objects. Using a PR2 robot equipped with BioTac sensors the group trained a robot to perform a set of exploratory movements while collecting haptic information from the BioTacs. In a separate experiment, the same objects were presented to human subjects who labeled them with common haptic adjectives such as (i.e. “fuzzy”, “squishy”, “smooth”, etc. – a total of 34 adjectives were used in the study). The robot was then trained with this data and learned how to associate the measured tactile information with the tactile adjectives. When presented with new objects the robot performed amazingly well and was able to correctly identify the properties of the new object.
The work, which was presented this year at ICRA, won the award for best paper in cognitive robotics. We would like to congratulate our friends and colleagues at U Penn and Berkeley for their great work! It has been a pleasure to have had the opportunity to work with your team the past few years and we’re excited to see such interesting work being done!
The original press release from IEEE spectrum can be found here.
The announcement from the University of Pennsylvania can be found here.
Research Article: V. Chu, I. McMahon, L. Riano, C. G. McDonald, Q. He, J. M. Perez-Tejada, M. Arrigo, N. Fitter, J. C. Nappo, T. Darrell, and K. J. Kuchenbecker, “Using Robotic Exploratory Procedures to Learn the Meaning of Haptic Adjectives,” ICRA, 2013. Download.