Konstantinova, J; Cotugno, G; Stilli, A; Noh, Y; Althoefer, K; (2017) Object Classification Using Hybrid Fiber Optical Force/Proximity Sensor. In: Proceedings of 16th IEEE Sensors 2017. (pp. pp. 543-545). IEEE: Glasgow, UK.

Preview

Text Stilli_Konstantinov Object Classification Using 2017 Accepted(redacted).pdf - Accepted Version Download (466kB) | Preview

Abstract

Intelligent perception to determine the physical interaction between robotic hands and the environment is a topic of great interest for the sensing and robotics communities. Sensor information on object stiffness and associated object deformation is essential to plan and execute stable and tight grasps. This paper proposes a novel robot-finger-integrated perception sensor to estimate the physical interaction with objects. The sensing principle of our combined force and proximity sensor is based on light intensity modulation, involving fiber optics technology to measure the proximity between robot fingers and object during approach as well as to detect normal and lateral forces. In order to distinguish between hard and soft (deformable) objects a Support Vector Machine (SVM) is employed to classify the handled object's stiffness based on the measured force and proximity data. The classifier does not require prior knowledge of the objects and achieves 87% classification accuracy on a set of household objects with different mechanical characteristics.

Download activity - last month

Export as XMLJSONCSV

Download activity - last 12 months

Export as CSVJSONXML

Downloads by country - last 12 months

Export as CSVJSONXML

Archive Staff Only

View Item