 A lot of people ask the question, why is robotics hard? We see a lot of demonstrations of systems, we see any number of YouTube videos showing self-driving cars, flying drones, quadruped robots, doing surprisingly complicated things, and they wonder how far are we from just having general, all-purpose, arbitrary, artificial intelligence and robots that can do everything from act alongside a warfighter to clean our kitchen and pick up everybody's kids' clothes off the floor. And the reality is that we are still quite a long way from realizing the dream of general purpose, artificial intelligence, especially in the physical world. The evolution of the field during the lifetime of the CTA evolved rapidly. And part of the evolution is that initially 10, 20 years ago, we were just trying to make those pieces work at all. Now we have evolved to a point where we can do this pretty well, except to do this now, we require an infrastructure that is not sustainable. So the research has shifted from being able to do any of those individual things, to now being able to do those things which we know how to do in a way that is consistent with actual operations. The most challenging question is exactly this point on how can we get robots to learn from experience and to kind of interact with the world and try certain things and then use this knowledge to improve their behavior over time. As we're getting into robotics, autonomous systems, artificial intelligence, these are things where we don't even know how to define what the problem is. We essentially need to understand the space and learn about what a solution even looks like together. And we need to build the biggest base of people in order to get at that problem. An advantage for the U.S. Army to partner with academia is the opportunity for them to train the next generations of scientists and engineers that will be leading the development, the design, and also the operation of these new technologies. We have lots of leading institutions in robotics that are involved with us in RCTA in particular. We basically get world experts in robotics thinking about our problems and giving their perspective and offering solutions. And it gives us access to a large pool of people, you know, excited, passionate students that are, you know, willing to sit with us in the field at integration exercises and suffer through the challenges of getting robots to work in the real world. We need these really diverse groups of people to be able to solve these problems because there's a diverse set of problems that we need to solve. And one of the things that we get in a program like the Robotics CTA is we quite honestly get the best researchers and the best people for working on these problems across the country. Not all of these things have gotten to a stage where they're sufficiently mature that they could be considered to be a candidate to go to a fielded system, but we made progress in all those directions and that's only in perception. The University of Pennsylvania has been involved in multiple projects, both in manipulation and also in navigation. Throughout the project we have produced many basic research results in both the mobility and manipulation part, as well as the perception. You can actually draw a line of some research which we did here, like with the detection and tracking with laser scanners, that was then used in the DARPA Grand Challenge and out of that then came Google autonomous driving and so on. But you have to do that fundamental research and there's still so much more to do. I think we made some good progress on the project, especially in the context of manipulation and a robot picking up objects, predicting how to move objects in the world, but of course this is really just the beginning and it's a long-term research endeavor and there's still a lot of work to be done and we're still pretty far away from anything that is at the level of kind of human intuitive understanding. I always envisioned the next phase of this RCTA would be having a lot of different robots, maybe even physically different robots collaborate together. We need to make AI far more robust. We need to make machine learning much more robust. We need to make it operate with far less data. I am not satisfied with where we are right now and neither should any of us be. There is a north star of having these robots work with and around our war fighters and we're not there yet, but we have the momentum, we have the ramp to be able to get there.