6 S's of Capable Home Robots

Well, considering the results of the Flight Telerobotic Servicer program, no, NASA couldn't.

John Nagle

--snip--

Gordon,

Think about the early dishwashing appliances (the term "dishwashers" could be considered ambiguous here ). I didn't own one, but a lot of people bought and used them in spite of having to pre- rinse, rack, and scrub the really sticky spots first, and kept using them even when they left egg on forks, warped or melted plastic dishes, and occasionally broke glassware.

I can't believe that _all_ of those early adopters were simply enthralled by the technology or concerned with impressing their neighbors no matter what the cost. So... how good _does_ a "mechanized assistant" have to be to begin selling?

(I should note that I still wash my dishes by hand.)

Frank McKenney

-- It may be that your sole purpose in life is simply to serve as a warning to others.

-- Frank McKenney, McKenney Associates Richmond, Virginia / (804) 320-4887 Munged E-mail: frank uscore mckenney ayut mined spring dawt cahm (y'all)

I think this is still the case. I have a reasonably new dishwasher and I still can't put dishes, pots, and utensils in that have caked-on crud. OTOH, when operated properly they use less water and energy, kill off some extra germs (the water is hotter), and might make a 30 minute job a

-- Gordon

Obviously. But my point was a *design capable of*, leaving accuracy, dependability, and liability as asterisks to the main goal of creating a robot physically capable of doing the tasks to begin with.

Is that a fair measure? Absolutely. Most well-established industrial robots fail miserably when used outside the highly controlled environment they were designed for. Robots used in automotive cease to function, and can cause damage to themselves and the product they are making, if there is an error of just fractions of an inch in the assembly line. There is little room for error. This is an accepted limitation.

This is what I've been saying all along.

-- Gordon

My 2 cents

A lot of the problems for a CHR result from the unstructured environment. I think the only way to progress is to work on the environment at the same time as the robot. People spend very large sums of money on kitchens and such like - what is needed is the kitchen/house to be a workplace suitable for some kind of roboticisation.

Consider cleaning the floor. One problem for a logical Roomba is navigation - now supose that all furniture/walls are marked with infra red visible bar codes - a simple scanner and not so simple software can build up a map. Similarly, if flexes must lie on the floor, make sure they are visible to that scanner.

A kitchen is a much more challenging place. One problem is that a robot takes up space, tends to either be in the way or falling over all the time

Imagine instead that you build into kitchen units and workspace a set of tracks. The robot hooks itself into those tracks, runs around to collect and deliver things. This could be quite a neat and compact arrangement, but also quite strong. ( Clearly you need to do all the obvious anti collision stuff )

You need cold storage that can be accessed by our railway line.

You store food in a limited range of containers, again with bar codes or whatever to identify type or location

I guess the point I'm making is that you need to consider the whole system, and if you do that I expect that no General Purpose Capable Home Robot will ever be built - rather you will have a home with built in robotic technology.

Dave

In my house I'd say less than 10% of the walls are accessible; everything else is covered by furniture, bookshelves, cat stuff. I'm not sure how I'd get barcodes on by Barcalounger.

The ceiling is usually pretty good for navigation. Some commercial indoor navigation systems (speficically NorthStar, promoted by Evolution Robotics) works this way. It's used in the Rovio WiFi toy robot, put out by WoowWee.

I think barcodes are a good idea for warehouses and other environments where adding the labels won't hurt the decor, and where the space is more open.

-- Gordon

Perhaps what's needed is a hand-held barcoder that has an inkjet head the squirts infra-red opaque transparent ink as you roll the head over a surface. That way you could invisibly barcode lotsa things and if some were unreadable, tough, there's be enough others visible.

The ceiling doesn't contain movable obstacles - the floor does.

Clifford Heath.

The final goal is to accomplish some of the tasks on my list of 90+ things that a robot could do. This may mean "cheating" when necessary to aid the robot using the things you mentioned. We often think about robots helping humans but in some cases the humans need to help the robot in some manner to collectively accomplish the goals.

Dave:

I think you have the correct basic idea although we would probably differ on the details. What you are describing is what I call "meet in the middle" robotics. This is robotics where the environment has been adjusted to help the robot without making it too onerous for the humans who must cohabit the environment.

For example, consider the task of clearing the table. With arbitrary silverware, plates, etc, it is a pretty daunting task. If the plates and silverware have RFID tags embedded, it *may* be much easier. If the plates and silverware are further color coded, it *may* get easier still. Similar, strategy can be employed for sorting laundry into whites, lights, and colors.

The trick is to make sure that the environment modifications are not too expensive or obtrusive.

-Wayne

Um, "NorthStar", celestrial navigation... Not a new concept here.

Obstacles on floors aren't used for navigation because they are not reliable sources of telling you where you are. You steer around them, but you don't calculate your position with them.

-- Gordon

You misunderstood. My point is that navigation is more than positioning. If you use enough markers around the walls at floor level you can help yourself identify obstacles when you *don't* see one you expect, because it's blocked. Two birds, one stone. All you need is a continuous frieze of (possibly invisible, IR) barcode around the walls.

Got it. Agreed.

(Though I think this is more practical in non-home settings where, in my opinion, is the more progressive setting for robotics use. This is an adoption problem, not one of technology. Example: when was the last time you saw an automated floor sweeper at McDonald's? That's where they're needed - not because McDonald's is dirtier than a home, just that it would be the natural place for a robotic floor sweeper. Driving industry to adopt these things takes a lot of time, money, and effort, but it's a significantly more profitable market than the home. Certainly iRobot is working toward this, as should other people. THESE types of applications are where we'll find the next Bill Gates...)

-- Gordon