How to predict dropped motion samples in haptic impedance devices

摘要

Stability and transparency have been always the most challenging concerns in haptic tele-operation. These issues are both impaired by any sample drop or transmission delay caused by the lack of reliability for the network channel, through which the two sides of tele-operation are communicating. Despite the contemporary high-speed data transmission schemes, this obstacle is still the most alarming one, specifically in IP-networks which provide packet-switched hence loosely-coupled connections. A common solution to this problem is to predict dropped or lost samples in either the impedance (master) or the admittance (slave) side of a tele-operation system. This research extends the existing methodology for this approach supported by an appropriate buffering technique within the impedance station. However, the admittance side is modeled by a remote virtual reality environment, which is connected through a WiFi IP network. The presented work takes several prediction algorithms into analogy in terms of the PSNR error metric. Investigation of the results recorded by an experimental study completed with human subjects has led to the most compliant predication method for the force-feedback data stream.