The main objective of the ADAS project is to develop an autonomous system for the identification and disassembly of electromechanical products such as TV sets or computer monitors.
ADAS aims to combine these elements to form a disassembly line for the dismantling of different devices of a product family (eg TV monitors). This line will be designed to be as flexible as is required to enable the identification of and adaptation to different product types.
It is therefore highly desirable that the sensor can detect the 3D co-ordinates of monitors without the restriction to diffusely scattering surfaces of smooth curvature. Also, for the objects of interest (diameter between 0.5 and 1 m), the required resolution that allows the identification of small components such as screws must be better than 1 mm. The use of the chirped laser radar technique offers a promising way to overcome these well-known restrictions.
As the disassembly process requires image-processing rates of higher than one 3D image per second, a fast image-processing technique is required. Neural networks have proven suitable for a number of image-processing tasks, but fast neural network implementations are necessary in order to perform online image processing. A trainable neural network PC expansion board with supporting software will be developed within the framework of this project. The implementation of disassembly processes will also require a graphical interface in order to enable users to develop strategies easily on their own.
Among the results to be expected are:
- strategies for the disassembly process
- integration of a sensor for fast 3D shape measurement
- development of fast image-processing software and hardware for shape recognition
- definition of a database for various electromechanical appliances
- development and implementation of disassembly tools.
CV1 2TT Coventry