Service Communautaire d'Information sur la Recherche et le Développement - CORDIS

Neural network-based robots for disassembly and recycling of automotive products

The integrated result of the Neurobot project consortium demonstrates the feasibility of a new generation of industrial robot systems, which are able to carry out automotive disassembly operations autonomously irrespective of car brand and type. These robots operate more precisely and more cost effectively, using neural network concepts. Instead of conventional programming, automated path planning and neural network vision is used to co-ordinate the robot actions. A multisensor system, consisting of non-tactile sensors transmits data about 3-dimensional position and material composition of automotive components to the task planning network. By means of automotive reference data and disassembly strategies the disassembly of relevant components is determined. Accurate movement of the robot as well as correction of errors and collision avoidance are controlled by means of the task planning network with the help of kinematic and dynamic networks. Special tools were developed such as a car handling unit, tyre removal gripper, windscreen cutter, glass crusher and battery removal gripper. The technology of the NEUROBOT system can be used partly on short term to assist in improvements in manual dismantling techniques. The application of a fully integrated system will need further development. With some necessary peripheral changes, this autonomous robot system will also be able to analyse, disassemble and sort other environmental harmful waste goods in order to feed them into the recycling process. The advanced technology is further applicable in a wide range of other complex manufacturing processes and for product handling, which were difficult to automate before.

Reported by

NEDCAR
5700 MC Helmond
Netherlands