Markus Vincze
Institute of Flexible Automation Technical University Vienna Gufihausstr. 25-29/361
A-1040 Vienna, AUSTRIA
Herbert Hufnagl
Institute of Flexible Automation, Technical University of Vienna Vienna, Austria
Karl M. Filz
Institute of Flexible Automation Technical University Vienna Gufihausstr. 25-29/361
A-1040 Vienna, AUSTRIA
P. Krautgartner
Institute of Flexible Automation Technical University Vienna Gufihausstr. 25-29/361
A-1040 Vienna, AUSTRIA
S. Spiess
Institute of Flexible Automation Technical University Vienna Gufihausstr. 25-29/361
A-1040 Vienna, AUSTRIA
Abstrakt
The development of artificial production systems reduced the need for perception to an absolute minimum. The lack of perception limits the flexibility of present flexible and integrated manufacturing systems. We claim that industrial production requires perception systems with different specifications to the human systems. Whereas humans are equipped to orient in arbitrary environments industrial tasks exhibit a certain degree of repetition and order. While order aids the general orientation, repetition calls for increased velocities to produce at a reasonable price. Particularly for the later requirements the human eye is hardly suitable, since it is not capable to observe or supervise automated processes, for example, the assembly of small parts. As conclusion we argue that industrial production requires two types of sensors, (1) very high speed sensors tailored to the task for processes such as part manufacture, part inspection, or part placement, and (2) versatile sensors, possibly with lower frame rates but a high degree of "intelligence" for processes such as part acquisition and recovery from errors. We outline the specifications for these types of perception systems and demonstrate the feasibility with example installations.