A strategy for sensors and rules in flexible robotic assembly

Abstract A suitable description for the problem- domain of assembly is a necessary prerequisite for practical error recovery. This article presents an alternative approach to flexible assembly involving sensory robotics by describing the assembly process itself in a framework which identifies the success of the intended operation. A rule is assigned to each of the sequential sub-processes in order to achieve each particular sub-goal state. Thus the use of artificial intelligence techniques provides a useful aid in a potentially complex situation of information processing. Error detection is achieved by receiving sensor values which deviate from the expected ones within the scope of a rule. By allowing the system to learn from its mistakes, costly pre-programming efforts are avoided and the cell is able to cope with unforeseen circumstances.