Challenges

AIMaReM Challenges and Approaches

CHALLENGE: Fast scanning times and large work pieces create very large data streams:

APPROACH: Compressive imaging and sampling, feature selection and classification in the compressed domain.

CHALLENGE: Online processing is required for active feedback in the process control loop

APPROACH: Sequential Monte Carlo – particle filtering, sequential compression, intelligent background removal - cointegration.

CHALLENGE: Classification and diagnostics under uncertainty including environment

APPROACH: Bayesian graphical models, info-gap uncertainty, data fusion.

CHALLENGE: New feature extraction required for NDE data streams and vision systems

APPROACH: high data rate capability for collection and processing.
Common CUDA approach for high data rate processing.

CHALLENGE: New software tools required for manufacturing automation

APPROACH: Vision and environment extensions to RoboNDT software.
Mobile robot capability for parts transport, fixturing and integration with inspection.

KEY ACTIVITY is combining and automating inspection techniques & data handling to deliver “smart” inspection – this includes the full process - part transport and fixturing is part of the problem space – new adaptive robot control systems are required.

ADDRESS
University of Strathclyde,
Electronic & Electrical Engineering,
204 George Street,
Glasgow, UK, G1 1 XW

CONTACTS
Email:
s.g.pierce@strath.ac.uk
k.worden@sheffield.ac.uk

LINKS