Positioning challenges in reconfigurable semi-autonomous robotic NDE inspection

This paper describes work conducted into mobile, wireless, semi-autonomous NDE inspection robots developed at The University of Strathclyde as part of the UK Research Centre for Non Destructive Evaluation (RCNDE). The inspection vehicles can incorporate a number of different NDE payloads including ultrasonic, eddy current, visual and magnetic based payloads, and have been developed to try and improve NDE inspection techniques in challenging inspection areas (for example oil, gas, and nuclear structures). A significant research challenge remains in the accurate positioning and guidance of such vehicles for real inspection tasks. Employing both relative and absolute position measurements, we discuss a number of approaches to position estimation including Kalman and particle filtering. Using probabilistic approaches enables a common mathematical framework to be employed for both positioning and data fusion from different NDE sensors. In this fashion the uncertainties in both position and defect identification and classification can be dealt with using a consistent approach. A number of practical constraints and considerations to different precision positioning techniques are discussed, along with NDE applications and the potential for improved inspection capabilities by utilising the inherent reconfigurable capabilities of the inspection vehicles.