TACTILE SENSING WITH COMPLIANT STRUCTURES FOR HUMAN-ROBOT INTERACTION
MetadataShow full item record
This dissertation presents the research on tactile sensing with compliant structures towards human-robot interaction. It would be beneficial for robots working collaboratively with humans to be soft or padded and have compliant tactile sensing skins over the padding. To allow the robots to interact with humans via touch effectively and safely and to detect tactile stimuli in an unstructured environment, new tactile sensing concepts are needed that can detect a wide range of potential interactions and sense over an area. However, most highly sensitive tactile sensors are unable to cover the forces involved in human contacts, which ranges from 1 newton to thousand newtons; to implement area sensing capabilities, there have been challenges in creating traditional sensing arrays, where the associated supporting electronics become more complex with an increasing number of sensing elements. This dissertation develops a novel multi-layer cutaneous tactile sensing architecture for enhanced sensitivity and range, and employs an imaging technique based on boundary measurements called electrical impedance tomography (EIT) to achieve area tactile sensing capabilities. The multi-layer cutaneous tactile sensing architecture, which consists of stretchable piezoresistive strain-sensing layers over foam padding layers of different stiffness, allows for both sufficient sensitivity and an extended force range for human contacts. The role that the padding layer plays when placed under a stretchable sensing layer was investigated, and it was discovered that the padding layer magnifies the sensor signal under indentation compared to that obtained without padding layers. The roles of the multi-layer foams were investigated by changing stiffness and thickness, which allows tailoring the response of multi-layer architectures for different applications. To achieve both extended force range and distributed sensing, EIT technique was employed with the multi-layer sensing architecture. Machine and human touch were conducted on the developed multi-layer sensing system, revealing that the second sensing skin is required to detect the large variability in human touch. Although widely applied in the medical field for functional imaging, EIT applied in tactile sensing faces different challenges, such as unknown number and region of tactile stimuli. Current EIT tactile sensors have focused on qualitative demonstration. This dissertation aims at achieving quantitative information from piezoresistive EIT tactile sensors, by investigating spatial performance and the effect of sensor’s conductivity. A spatial correction method was developed for obtaining consistent spatial information, which was validated by both simulation and experiments from our stretchable piezoresistive EIT sensor with an underlying padding layer.