Loss of communication between humans and machine leads to a loss of control. An Air Force predator drone crashed in Afghanistan in 2010, and it is being investigated. NASA lost a Martian probe in 2006 due to a loss of communication. Artificial intelligence does not yet fully address this area of lost communication. The on-board computers, in a loss of communication event, must ask and answer itself the question, "What do I do now?" In the predator case, if the computers were programmed to return to home base if communication was lost, perhaps it would not have crashed. In the NASA case, the on-board computers could have been programmed to simply orbit Mars until communication was restored. This area makes a good thesis idea, since programming has to account for the contingency of lost communication.
Human beings make decisions often based on incomplete data and readjust the decision according to fresh data coming in. For example, you have to drive to work. It starts to rain lightly, so you put the windshield wipers on "slow." It starts to rain harder, so you put the wipers on "fast." It rains even harder and the road becomes flooded, so you pull over. These three simple decisions actually can be broken up into hundreds of micro-decisions, based upon you prior learning. One such micro-decision was at exactly what time do you turn the wipers on. Artificial intelligence systems could be programmed to act upon new data and readjust the output based upon fresh incoming data. For example, a NASA probe is sent to Jupiter. Its on-board radar senses an asteroid straight ahead and readjusts the flight angle to avoid the asteroid. Other examples can be delved into, such as an auto pilot that senses a side wind or an intelligent car radar that senses another car while driving through fog.
Fuzzy Logic is a philosophy that combines the concepts of loss of communication, acting on incomplete data and acting on "gray area" data with no clear-cut yes or no answer. A robotic system must act on all of these ambiguous inputs with a definite output. According to the Seattle Robotics Society, professor Lotfi Zadeh at the University of California at Berkley believed that humans do not need precise, accurate inputs but are capable of highly adaptive control outputs. In the predator example, suppose communication was lost. If the on-board computer had a Fuzzy Logic system, it could continue to gather data, follow the terrain by varying its altitude, complete its mission and fly back to home base. On the way back, if it encountered a head wind or side wind, it could readjust its flight path accordingly. Fuzzy Logic addresses all of the variables involved. In your thesis, discuss what needs to be done so a control system uses Fuzzy Logic as its input and develops successful outputs.