Our research group here at USC works on a group of tasks around design theory, systems engineering, self-organizing systems, and machine learning. Part of the group is researching engineering methods and creativity (design theory work). Some of the group works on self-organizing system. The rest (myself included) are working on machine learning, cognitive computing, and collision avoidance strategies.
The collision avoidance problem is a topic that has been investigated for millennia, however, it was in the 1840's that the British government decided that the amount of shipping that was being lost due to collisions at sea needed to be resolved. From this point on many articles and root cause analysis investigations have taken place for shipping, automobiles, and aircraft. While the incidents of collisions have gone down, there is still an average of one collision that occurs each day at sea alone. And each of these collisions at sea have an average economic impact of $500MM.
The issue as we in our research group understands it is that there are non-linear and non-intuitive effects that lead to collisions. To assist in this our group is looking at ways that cognitive computing, can assist us in understanding what are the fundamental aspects that go into collision avoidance systems. This would be on the natural language processing part that cognitive computing can bring. The second part of what we are looking for is on the numeric and state driven risk analysis of collision scenarios. What are the "moves" that one should make at any given time that would allow the user to achieve the goals that are needed (safely bringing the cargo to port, landing the airplane, going to the mall in the car, etc.) without adding risk to the process.
What our immediate group is looking for is a way to take these data and functions and create guidance solutions to minimize the risk using cognitive computing. We also want to add more complexity over time for these models to include non-mechanical and non-numeric data to evaluate risk. We want to develop ways to provide the human crew and decision makers with the right set of information at any time given the system state (not just position) and data available. On the design side we have looked at how to use the expertise accrued across an organization for engineering design using cognitive computing methods. Finally, our group is not a computer science or computer engineering group; we are a group of mechanical, aerospace, and systems engineers. Because of this we would like documentation on how to create and deploy software for the Intel platform.