Robots exist that have worked faultlessly for years, perfected by an iterative process that fixes the most frequent failures, revealing successively rarer problems that are corrected in turn. Unfortunately, that kind of reliability has been achieved only for prearranged routes. An insectlike 10 MIPS is just enough to track a few handpicked landmarks on each segment of a robot’s path. Such robots are easily confused by minor surprises such as shifted bar codes or blocked corridors (not unlike ants thrown off a scent trail or a moth that has mistaken a streetlight for the moon).
A Sense of Space
Robots that chart their own routes emerged from laboratories worldwide in the mid-1990s, as microprocessors reached 100 MIPS. Most build two-dimensional maps from sonar or laser rangefinder scans to locate and route themselves, and the best seem able to navigate office hallways for days before becoming disoriented. Of course, they still fall far short of the six-month commercial criterion. Too often different locations in the coarse maps resemble one another. Conversely, the same location, scanned at different heights, looks different, or small obstacles or awkward protrusions are overlooked. But sensors, computers and techniques are improving, and success is in sight.
My efforts are in the race. In the 1980s at Carnegie Mellon we devised a way to distill large amounts of noisy sensor data into reliable maps by accumulating statistical evidence of emptiness or occupancy in each cell of a grid representing the surroundings. The approach worked well in two dimensions and still guides many of the robots described above.
Three-dimensional maps, 1,000 times richer, promised to be much better but for years seemed computationally out of reach. In 1992 we used economies of scale and other tricks to reduce the computational costs of three-dimensional maps 100-fold. Continued research led us to found a company, Seegrid, that sold its first dozen robots by late 2007. These are load-pulling warehouse and factory “tugger” robots that, on command, autonomously follow routes learned in a single human-guided walk-through. They navigate by three-dimensionally grid-mapping their route, as seen through four wide-angle stereoscopic cameras mounted on a “head,” and require no guide wires or other navigational markers.
Robot, Version 1.0
In 2008 desktop PCs offer more than 10,000 MIPS. Seegrid tuggers, using slightly older processors doing about 5,000 MIPS, distill about one visual “glimpse” per second. A few thousand visually distinctive patches in the surroundings are selected in each glimpse, and their 3-D positions are statistically estimated. When the machine is learning a new route, these 3-D patches are merged into a chain of 3-D grid maps describing a 30-meter “tunnel” around the route. When the tugger is automatically retracing a taught path, the patches are compared with the stored grid maps. With many thousands of 3-D fuzzy patches weighed statistically by a so-called sensor model, which is trained offline using calibrated example routes, the system is remarkably tolerant of poor sight, changes in lighting, movement of objects, mechanical inaccuracies and other perturbations.
Seegrid’s computers, perception programs and end products are being rapidly improved and will gain new functionalities such as the ability to find, pick up and drop loads. The potential market for materials-handling automation is large, but most of it has been inaccessible to older approaches involving buried guide wires or other path markers, which require extensive planning and installation costs and create inflexible routes. Vision-guided robots, on the other hand, can be easily installed and rerouted.
Fast Replay
Plans are afoot to improve, extend and miniaturize our techniques so that they can be used in other applications. On the short list are consumer robot vacuum cleaners. Externally these may resemble the widely available Roomba machines from iRobot. The Roomba, however, is a simple beast that moves randomly, senses only its immediate obstacles and can get trapped in clutter. A Seegrid robot would see, explore and map its premises and would run unattended, with a cleaning schedule minimizing owner disturbances. It would remember its recharging locations, allowing for frequent recharges to run a powerful vacuum motor, and also would be able to frequently empty its dust load into a larger container.
See what we're tweeting about
40 Comments
Add CommentI love this.
Reply | Report Abuse | Link to thisI only hope my father lives long enough to see this.
"Ultimately, though, it is likely that our descendants will cease to work in the sense that we do now. They will probably occupy their days with a variety of social, recreational and artistic pursuits, not unlike today’s comfortable retirees or the wealthy leisure classes."
Reply | Report Abuse | Link to thisIf the situation in Saudi Arabia is an example, a large number of people with no responsibility in life isn't such a great thing. We'll have youths roaming the streets committing acts of terror on unsuspecting robots... until the robots finally rise up against their masters.
Back in the sixties, people started automating jobs. The unions got worried, but were told 'You are now entering the age of leisure...computers and robots will mean less work for you'. Of course, as we now know, robots meant job losses. And as computers are pretty dumb, our jobs became even dummer to conform! There are now 970,000,000 starving people in the world because of automation in Western agriculture. Where will it all end?
Reply | Report Abuse | Link to thisI don't beleive that robots can perform anywhere near as well as a lizard. Take, for example, the case of a much more 'lowly' beetle. Its brain is minute, yet it can walk, run, avoid any object, climb any vegetation, swim, dive, burrow, fly, navigate, calculate intersecting trajectories etc, and that using all of 'our' senses, not to forget feed, breed, even when injured. The problem with computers, is that lack the massively parrallel structure of nervous systems. Our computers lack adequate natural selection.
Reply | Report Abuse | Link to thisNot sure. Yes, we really do need much greater processing speed (or distributed processing, as our brains are relatively slow paced processes in comparison).
Reply | Report Abuse | Link to thisBut I think the real deficiency will always be how we "program" them.
If we could do that today, even with sub-par processing speed, we could simulate a much slower than real-time process. Maybe this has already been done and I am just out of the loop.
i dont think computer can act like a human,especially they can have a brain thing.the data is different of electronic impulse of our brain.
Reply | Report Abuse | Link to thisYeah, right. There will be no robots with super brains. That requires an extremely high energy society. And that society in which we live is at the apogee of its existence. From here on in, the energy ride is down. Peal Oil. Peak Uranium. Peak Metals. Peak Earth. Technotopians will revile me, will pull out their hobby horse, their favorite solution, which will inevitably be remarkably bereft of any physics knowledge, especially of that basic harsh truth: We live on a sphere. Spheres are finite. But, unbridled technoconsumption and technophilia sells magazines. So, have at it. You'll be lucky to have the Internet in ten years.
Reply | Report Abuse | Link to this"Ultimately, though, it is likely that our descendants will cease to work in the sense that we do now. "
Reply | Report Abuse | Link to thisRight, assuming your robots do pick up all the work. But then "They will probably occupy their days with a variety of social, recreational and artistic pursuits, not unlike todays comfortable retirees or the wealthy leisure classes." hahaha, how naive.
We have enough experience there; people with no work are people with no sustenance, they are those starving and living with less than a dollar a day, see them all over the world. Will the robot-manufacturing corporations feed the masses for free? Is that what's on your mind? Dude, get out of your lab more often, replacing workers for machines ultimately kills people. Having those nice smart robots won't change the nature of it.
People not working in Artificial General Intelligence cant see that superintelligence is nearly here, because they dont know what methods we are cracking to deliver it.
Reply | Report Abuse | Link to thisAlthough Kurzweilian estimates are for 2020-29, and Vingean for 2005-30, it could be much earlier and some people like Minsky and MacCarthy have believed it possible since the 486 was on line.
I've worked in a human robots factory & it's possible robots could be leaped over to Superintelligence.
Cheers
eldras
london A.I. Club
A lawyer may be better at everything his assistant does than the assistant, but that doesn't mean he's going to waste his time doing it. It won't be any different when robots are better at everything than humans. Opportunity cost works the same for both robots and meatbots.
Reply | Report Abuse | Link to thischronic unemployment is from regulations like minimum wage laws driving the cost of labor above it's value, not automation.
Most methods to AI today are hard coded rules and really aren't intelligent because they don't understand the "why"(factors) of anything they do. I wrote a lot about it because I'm trying to develop human-like AI to do just what this article talks about. But most of the world is going in the wrong direction and even with a computer the size of earth, it still wouldn't be intelligent with today's AI algorithms.
Reply | Report Abuse | Link to thisIf you'd like to learn more about what human-like AI takes, check out my website: http://www.PracticalAi.org
While he may be right about the time line for robots he may be wrong about the time line for AI. There are groups of people that work in collaboration on such things. One example is PracticalAi.org
Reply | Report Abuse | Link to thisI think the biggest problem with AI programming right now is that it relies on a static view of the world. Our own minds do not work this way. For example when we see something; we "See" an item we have seen before the closely resembles the actual item in front of us. Often our first impression is wrong but as we see more of the item in front of us we continually refine what we think we see as we get more detail. We then acquire as much detail as we need to interact (or avoid) the item. Then if the item was not important we forget more and more details about it until it is either represented by a single symbol or we forget about it all together.
Reply | Report Abuse | Link to thisPeople programming walking robots finally go this concept but the people working on the AI can't seem to wrap their heads around it.
As for navigation the robot needs to identify the objects to properly move around them. That is a wall: it does not move I can put WALL SYMBOL and boundary and not examine that object next time I pass this way. That is a chair: chairs move a lot (and may be in motion): I will need to look for that object (and gather ONLY enough information to identify that chair) to place where it is when I come through next time. That is a person or other robot or animal: it is in constant motion once I have identified the static objects in my path I need to track these objects and try to predict their motion.
People are messy thinkers and movers we are in a constant state of balance and changing our positions and perceptions constantly. Also WE bump into each other, make wrong turns, misjudge distances, misinterpret what we see all the time. We simply adjust as we go.
The bottom line: Life is dynamic not static.
As for the the argument that automation takes jobs and leads to violence through boredom. These are SOCIAL issues that require we work on improving these area through social developement.
Reply | Report Abuse | Link to thisWe SHOULD be living in a high-wealth, low-labor sociely NOW!
We have the technology AND the resources to provide everyone WORLDWIDE with a very high standard of living and reduce the amount of back-breaking labor to a minimum.
We should be working four, four hour days a week and enjoying a middle class life style (house, car, food, medical 4 weeks vacation time etc.) but the combination of waste and greed at the highest levels has robbed us of this.
Ahh, the much anticipated (and feared) singularity. The point at which humans design an intelligence smarter than themselves, essentially negating the need to design anything further. I tend to side with those that think the singularity could occur in the next 50 years. This could result in a wonderous golden-age alluded to by the author, or a nightmarish situation where the human race becomes unnecessary altogether. Unfortunately, there is not a reliable way to predict the actions of such an intelligence because it will be, by definition, beyond our understanding. Here's hoping for the best...
Reply | Report Abuse | Link to thisI am skeptic about it.
Reply | Report Abuse | Link to thisFirst, If we move to a competence for resources with smarter beings (aliens, robots or ninja turtles for the case), we will inevitably loose, that is the basis of today human vs human wars, and probably will be for future wars.
Second, the same thing was said about nuke power, that it would become so cheap that electricity companies wouldn't bill you because it would be more expensive to gather the money than to give electricity for free.
And third, there will be a point where the "uncanny valley" will get the best of us, so we will have boxy typed robots with a limited intellect (like S.A.R.A.H. from Eureka series) that can understand and serve their owners but are too dumb to outsmart them, don't look like humans and are too naive to learn there are bad persons.
AI is to computer science what porn is to visual arts. It's almost always about to work but never does. We're not anywhere close to figuring this out. Even babies require several years of training before they can clean, dust, and what not.
Reply | Report Abuse | Link to thisWhy do robots have to think and act like us to have intelligence? They are different from us in every way. They will never take over all our jobs, only the basic one any was e.g. agriculture and industrial produce. Creativity and theoretical thinking will always be the specialty of humans. I think by 2050 humans' job in this world will thinking mostly e.g. invention of new ideas.
Reply | Report Abuse | Link to thisbeware the age of the robot, more humans out of work, and who knows skynet may become a reality.
Reply | Report Abuse | Link to thisRead "Shadows of the Mind" by Roger Penrose. No computer, even if infinitely powerful, can ever hope to hold a candle to the conscious human brain. And consciousness is probably related to quantum phenomena. So much for Artificial Intelligence.
Reply | Report Abuse | Link to thisThere are two robots on Mars called "Spirit" and "Opportunity". Squyres complains that it takes a robot all day to do what a human can do in fifteen minutes. Enough people agree with him to NASA scheduling manned space flight again, which I believe is foolishly expensive. How far away are we from designing a robot that could do the jobs that an astronaut would do on Mars in the same amount of time?
Reply | Report Abuse | Link to thisYeah, they're toys in 2009, we're slaves to them in 2109. Robots with intelligence of a human being are bad because...
Reply | Report Abuse | Link to this(a.)Too many robots would significantly lower the amount of jobs with a growing population which equals poverty. (b.) The more intelligent robots get the more inclined they would be to take over and annihilate all humans.
Yeah, they're toys in 2009, we're slaves to them in 2109. Robots with intelligence of a human being are bad because...
Reply | Report Abuse | Link to this(a.)Too many robots would significantly lower the amount of jobs with a growing population which equals poverty. (b.) The more intelligent robots get the more inclined they would be to take over and annihilate all humans.
If the sharing of wealth is similar to today's patterns, the world will probably crash. If the wealth is shared, population trends will accelerate and the world will probably crash. We have plenty of human intelligence now, and the world's biological sustainability is going off an edge (the world is going to crash). Who the hell needs robots to take out our trash and drive our cars? In all probability the world's elite will have them pointing guns at us. By the way to get more of my transhuman perspective go to: http://www.manyone.net/MonSanc/
Reply | Report Abuse | Link to thisBy the way, the technical elite are genuinely myopic in their rosy scenarios.
If the sharing of wealth is similar to today's patterns, the world will probably crash. If the wealth is shared, population trends will accelerate and the world will probably crash. We have plenty of human intelligence now, and the world's biological sustainability is going off an edge (the world is going to crash). Who the hell needs robots to take out our trash and drive our cars? In all probability the world's elite will have them pointing guns at us. By the way to get more of my transhuman perspective go to: http://www.manyone.net/MonSanc/
Reply | Report Abuse | Link to thisBy the way, the technical elite are genuinely myopic in their rosy scenarios.
On March 16, 2006, The Globalist published an article by me suggests a very different time-frame and covers these issues from a rather more human (and humane) point of view. Titled "Will Japanese Robots Rule the World by 2020?", it is available at: http://www.theglobalist.com/StoryId.aspx?StoryId=5084
Reply | Report Abuse | Link to thisIt's logical to assume that if you want a robot to perform properly in the real world you should mimic nature. Start with a neural net based on an insects mind. If you choose a simple evolutionary chain of creatures that ultimatly led to the human brain then you have the ability with each successive model to upgrade in the evolutionary chain. I don't think using the serial computer model is going to work properly in the end. That's what I call the Microsoft approach, throwing lots of horsepower at a problem like they did with the Pentium chip. Two smaller fast chips can do more work than one big one. Neural nets are key to making robots work properly. Let Mother Nature teach you what she has learned over millions of years. There are some basic commands built into these nets for example it's the reason a foal can stand and walk just after it is born. Human's are preprogrammed to understand the human face. So you have to build some of these generalities into your basic robot brains. If you can do that then they will truly evolve. Commercially you can sell these preprogrammed add-ons. For example, this robot understands facial expressions and this add-on understands emotions.
Reply | Report Abuse | Link to thisOne other idea I thought of was to image the brain of a racoon as it is quite industrious for its size. Set up a virtual reality version of the brain in a computer and program it to behave the same way as the real one would work with virtual neurons firing. Sample the real thing and see if you can make the virtual version behave in the same way. Then you would have a kick-ass robot brain!
Ok lets think about this, back in the ancient times there were no robots and no need for them! Today were wasting our money and talent on these stupid things! Theyre only gonna take our jobs, make us lazy, and kill us.
Reply | Report Abuse | Link to thisits very much impressive.give some brief article.
Reply | Report Abuse | Link to thisThere seems to be a need to put all this computing power in every device,when robots could operate via wifi from a mainframe.One central computer could run many robots doing many different things.Each robot would be lighter not having to carry heavy computers,but only enough to to operate the machinery of the individual robot with the thinking being done by the mainframe.This would allow for human mind level computers much sooner,maybe less than a decade instead of 30 years.Unlike humans taking decades to learn,once one computer learns it can be passed on to one another like a recording without each mainframe having to learn from scratch.Using this method we may be only a few years away from a thinking machine by just hooking together a bunch of pc's and supercomputers and programing them with self learning programs instead of trying to build one huge computer.
Reply | Report Abuse | Link to thisWe already have many robots just in human form and that perhaps a lot more in some places than in others.
Reply | Report Abuse | Link to thisIf we're thinking of creating human-like machines, both hardware and software matter.
In regards to software, the computers of many humans when unused could each run a software to simulate the thinking and learning of a human-like machine using the many algorithms and even the Internet. This learning could be shared and collaborated with software on the other computers. The software could be programmed to be like academia, industry, professionals, and others and that each would generate their own knowledge and experience data which they would further share and collaborate with software on the other computers.
When I had started undergraduate studies at Georgia Tech about a decade ago, I had wanted to pursue aerial robotics because of interest in autonomous flying vehicles. A grad student of the top three ranked program in aerospace engineering had mentioned that I could pursue it from several disciplines and that didn't even have to pursue aerospace engineering. I pursued computer science but didn't pursue aerial robotics after that. Over the decade, I've learned on many things related to computer science from business to systems applications. Now, I'm trying for grad school admissions in quantitative computational finance. Just like the grad student mentioned, even quantitative computational finance knowledge and experience would help to pursue human-like machines.
artificial intelligence shall know things more than just moore's law, or brain researchers already degraded into computer chip researchers
Reply | Report Abuse | Link to thisi believe today's artificial intelligence researcher or commentators shall know more than just moore's law and a dumb clumsy brute force algorithm for searching, or brain science researcher already declined into computer chip researchers, obviously brain is superior to digital chip
Reply | Report Abuse | Link to thiswhen talking about its application, the key is not what they can do, the key is its social impact, their safety to human, and whether our society could get into utopia and dystopia.
i could assure you such a stuff is on the way, and it will be much earlier, the true limitation is not funding, but how many people having a die hard spirit to make it happen, who do that is more important than how much money is granted to do that
"it is a huge undertake by thousands of human programmers and computer alike"
Reply | Report Abuse | Link to thisthe truth is a robot software is smallest among all type of useful software at all, you could see no robot company startup have more than 50 people today, when talking about success, at least i could ensure, such thing will not happen, or happened in startup
okay, so this is just awesome? haha
Reply | Report Abuse | Link to thisthis is so awesomeeeeee
Reply | Report Abuse | Link to thisthis is gay
Reply | Report Abuse | Link to thishey nate
Reply | Report Abuse | Link to thisthis websites gayyy as he**
Reply | Report Abuse | Link to thisThere are plans on the drawing board for a Precongnizant computer capable of foreknowledge, mind reading and discerning activity in the future with a high degree of accuracy and correctness: the correctness, truthfulness and the ability to be precise and avoid errors. As the designer of the Blue Prints of this new process it uses a new type of computer software that combines code with imaging. This dual lobe process is also selfevolutionary allowing it to learn at exponential model rate. The system uses Sensory audio and visual which provides highly accurate, embedded speech/vision recognition solutions for both integrated circuits and embedded software platforms.
Reply | Report Abuse | Link to thisSenses with physiological capacities built within that provide inputs for perception and feed back while self evaluating gives rise to the future generation of robots. Is this possible, first man must conceive the idea, think of or imagine something, form the idea or concept in your mind. You must then invent or devise a plan for the invention then put it into action.
Ron Nussbeck