Sramana Mitra: I don’t see this happening in the food preparation domain as much as I see it happening in cleaning services. Lawn mowing is a good example, because it is a repetitive and physically exerting function. Those are good application areas. If I were thinking about consumer applications, what other applications would I be looking at in terms of launching a product to market that could be remotely as successful as Roomba?
Rich Mahoney: The number one request I hear is from people [is for a robot] to clean their kitchens, wash their dishes, clean toilets, do laundry, fold laundry, etc. These are very sought-after tasks for robots. I respect your comment whether food preparation is going to be viable or not. The thing for me about the market that is emerging is that I think it is important that the technology is acceptable and that the development community is really going to drive the solution space. I don’t agree or disagree with your comment about food preparation, but I think the same way we saw innovation in computing, we are going to see innovation in robotics. It is going to start from an entrepreneurial base, and it is going in a direction that is not even clear right now.
SM: The general proxy of what you are looking at in the domestic household robot is the one you would normally domestic helps performed, like cooking and cleaning, for example. That is where you would want robotics to intervene and make an impact. My reservation was more towards how realistic it is to imagine that those are entirely feasible. It is not that much of a repetitive task. Something that is very structured and highly constrained, like a vacuum cleaner’s or a lawnmower’s work is one thing, but unloading a dishwasher and deciding which dish goes where, for example, is much less constrained and much less of an algorithmic notion.
RM: It goes to that same point that the accessibility of the hardware would then allow entrepreneurs to create solutions to those situations. Maybe kitchens become robotically convenient and are structured in a way that allows the first class of robots to be used. All dishes could have little symbols attached so the robots would know which cabinet they belong to. There are lots of ways to handle that. People who are developing products are going to come up with all kinds of solutions for different environments.
SM: You have to think about it from the kitchen design point of view. You cannot retrofit a kitchen that is designed a certain way.
RM: I don’t see any issues in that sense. There are robots that will emerge that are general purpose and will be able to work in the kitchen without much retrofitting. Not knowing where to put a dish is not that much of a problem. You may have 500 items in your kitchen, and you have recorded somewhere where each one of those goes. You can also have a camera image of each one and have some kind of optic recognition. It is not that difficult of a problem. The software solution is not the problem here. It is the robust manipulation hardware of the robot that is really the challenge.
SM: Users will have to spend a fair amount of time setting it up. That is where I think adoption won’t go that well, because the user has to do a lot of setup.
RM: There are many ways to feed that information into a robot. Think about how much is available on your cell phone now. I don’t know if you ever went into Google Sketchup and looked at all the different items available there. There is a very large data set of known objects that are available now. I can think of several different methods that would allow the consumer to document everything in his or her kitchen without too much trouble. The software side of it is not the challenge, it is the manipulation hardware that is robust and can actually reach and manage all of that space in the kitchen.