 You have heard about 2D manufacturing technologies in the previous two talks. Let's talk about 3D, asking the question whether we can imagine designing machines that can design themselves and assembly lines that can self-reconfigure to synthesize the best shape for the product or for the manufacturing task at hand. If we could do this, we will transform manufacturing fundamentally because right now developing tools and robots is done in a customized way and this takes years and this has caused manufacturing to lag in automation. For example, the car manufacturers can automate 80% of their tasks and this is because they can take advantage of high volume and repetitive processes. But in contrast, the electronic industry automates only 10% of the tasks and the plane manufacturers only about 30% of the tasks and this is because they have products with short life cycle, low volumes, or very highly complex. So a natural question is how can we enable all manufacturing, all industries to achieve a high degree of automation independent of how short the product life cycle is, how low the volumes and how complex the products. And the opportunity is to use computation to begin automating the design and the fabrication of the tools and robots on the manufacturing line because that's where the fixed points are. So if we can automate the design of our tools, we will be able to eliminate the fixed points. Now every manufacturing process, every assembly line requires tools and processes. What if we can specify the function of what you need and have the right tool be synthesized on demand along with all the processes required to use it? And what if we could give our workers the ability to do that by themselves with very easy intuitive processes and when the product changes, the same worker using the same tools could push a button and reconfigure the assembly line. Recompute what is the correct tooling and what are the correct processes required. So how crazy is this idea? Well say you need a new manipulator that can perform a particular type of task. What if you could just type the function of that manipulator inside text editor and then have a computation engine designed behind the scenes what that manipulator would look like, then fabricate it, then create the computational infrastructure required to use it and the programming infrastructure required to program the device. This idea is adaptable to everyday life. So here is one of our users Alice who wants to automate many tasks in her home. Maybe she wants a playmate for her cat while she's at work. So equipped with a rough idea of what Alice wants, she can head to 24-hour robotics where she gets access to a user interface that is intuitive and it's very much like the interface we imagine for the workers in the factory. And when she's satisfied with the design, 24-hour robotics can print the design along with the programming environment required to use that design. And as a result of this intuitive process that starts with an idea and leads to a functioning product, the cat has a playmate. And so this idea behind using computation to generate robots on demand has the potential to democratize access to robotics and create a whole new user community that can share experiences. So how feasible is this? Well, say I want a robot to play chess with me. What if I type, I want a robot to play chess with me in my browser and my computer synthesizes the best robot shape along with all the intelligence required for this robot to play chess with me. And success, I get my robot and this robot can actually reason through a chess game with me. This robot can also play with a cat. We've created this robot in an automated way in a couple of hours for a cost of a few tens of dollars. And it's very exciting because the new manufacturing processes that we have developed start with 2D printing processes that have the built-in intelligence that enables us to take the printed sheets, put them in an oven, apply uniform heating to them and then bake them into the three-dimensional structure that we would like for our robots to have. So the big dilemma in today's manufacturing is how do we go from mass production to customized production? And I believe that by using computation and robots and by creating the infrastructure to create one robot for any task using functional specifications starting with the demand we can truly revolutionize manufacturing following the principle that if you can think it, you can build it. And for the discussion question I would like to talk with you about how would one robot for any task reshape manufacturing?