For many years robots have been programmed at manipulator or joint level without any real thought to the implementation of sensing until errors occur during program execution. For the control of complex, or multiple robot workcells, programming must be carried out at a higher level, taking into account the possibility of error occurrence. This requires the integration of decision information based on sensory data. Aspects of robotic workcell control are explored during this work with the object of integrating the results of sensor outputs to facilitate error recovery for the purposes of achieving completely autonomous operation. Network theory is used for the development of analysis techniques based on stochastic data. Object level programming is implemented using Markov chain theory to provide fully sensor integrated robot workcell control.