Thursday, October 21, 2004
Silicon Valley is unlikely to be "robot valley" in the US
More evidence that the tech center for the emerging robotics industry is unlikely to be in the focal region of the PC and Internet revolution - Silicon Valley. Proof is seen in an article in Mercury News reporting on the upcoming RoboNexus conference. Like other robot articles in the Silicon Valley press, the rise of the robots at the weekend trade show is treated like a report on a strange circus carnival. The surprising fact that robots exist is reported, coupled with happy re-assurances that nothing will happen for many years and we're a long way away from anything a Silicon Valley person would care about.
Why the soothing message of robotic irrelevance? If robots are really on the rise, Silicon Valley is in trouble. Clearly, the average Silicon Valley computer developer has no more knowledge of robotics than anyone else - bad news for this group ever "leading" the rise of the robots.
A few cool companies like Anybots (humanoids) and Team Overbot (DARPA Grand Challenge for robot-cars) are working in the Silicon Valley area. But they are islands in a sea of groups working on "more important" things (presumably another try at ITV and Linux) that are linear continuations of the last 20 years of computers and cyberspace. These dorks are slavering over Bluetooth in your pen or the latest crazy hiring questionnaire from Google (uncool, non-Matrix-lovin', non-digiteratti who don't read Wired magazine need not apply).
Of course, the detractors of robotics have a point. When the PC first appeared in the early 1980s, some assumed that the HERO-ic age of robotics had dawned, and hobbyists began creating simple robots while professionals put embedded microcontrollers into cars and microwaves. They're still doing this - often with the same 20-year old, 4MHz 1Kb memory microcontrollers they used then. The fun was all in the creation and understanding embedded systems - hardly a mass-market preoccupation. During the 1980s and 1990s, both research robots and hobby robots were the source of periodic press releases concerning slightly laughable machines with no connection to anything the non-specialist would care about. Meanwhile, Hollywood told us that "robots" were formerly unemployed SAG actors with grey makeup and a mean face.
Today, we're seeing a different world. The U.N. Economic Commission for Europe and the International Federation of Robotics just reported reported 607,000 automated domestic helpers were in use at the end of 2003, 67% of them purchased that year. Most of them - 570,000 - were robot lawnmowers. Sales of vacuum cleaning robots were reported as 37,000, (but this is false - iRobot has sold over 500,000 Roombas). According to the report, by 2008, some 4.1 mln domestic robots will be in use, the study said. Service, entertainment, and domestic robots will soon eclipse industrial robots, currently at about 1 million worldwide.
7-fold in 4 years...that's full-bore growth like the PC industry experienced in the 1980s, and much faster than the tepid growth of e-commerce (up about 20% in 2004) or Internet usage (it actually dropped during the first half of 2004).
And where is the investment? The Mercury article notes that VCs are "uninterested" in robotics at present - I suppose they're still waiting for the next company to try sending "movies over your cellphone" or another useless thing
...those were the days, my friend, we won't admit that they reached the end...(music)
I suspect that "faster" or "beter" hardware is not the reason for this spurt in robots. The iRobot Roomba uses the same components that a hobby robot does, but it does something useful. It is loaded with top-flight engineering and industrial design aimed at consumers. There's been a change in mindset about robotics, allowing useful products to appear.
Possibly, with the waning of cyberspace as a "hot thing" and the decidedly un-newsworthy spread of recent "big next things" such as wireless, there is at last space in the collective unconscious for a different computing paradigm. One thing that is clear - robotics is not just another kind of PC, even though robotics companies are stealing PC technology for their decidedly un-PC creations. Programmers of the virtual world will have a hard time making the jump - but a few will. With the success of a few platforms (e.g., Friendly Robotics lawnmowers, and home-bots in Korea and Japan, entertainment "super-toys" like the Sony Aibo) the industry is now waking up and realizing that the goal of building a robot is not to use a single 1980s microcontroller for a proud display of line-following.
PC robots using a standard personal computer OS like White Box Robotics have been attacked for the use of a sloppy, wasteful operating system like Windows. Who cares, except an embedded systems purist? If it works, use it - hijack it for your own purposes. The early personal computer industry hijacked microprocessors intended for other purposes. The PC-derived robot is one of several attempts to develop standard hardware and software platforms for consumer robotics and put a standardized "floor" under at least one class of consumer-bot. The point is to get something out there that works, that non-embedded programmers can use, without constantly re-inventing the wheel. As the industry grows, it will be possible for hardware developers to make chips and other hardware specifically for robots, and we'll see the real explosion.
I suspect that robotics will soon develop its own OS family (not based on standard embedded OSes, either) but for now, creating first-generations Robots That Move is the priority. Hardware that can do more than do more than crawl across a lab floor in 4 hours is a priority. Cheap legs from Anybots and/or balancing-wheel bots like the Segway RMP platform are what's needed for Silicon Valley to finally wake up and smell the robot.
More evidence that the tech center for the emerging robotics industry is unlikely to be in the focal region of the PC and Internet revolution - Silicon Valley. Proof is seen in an article in Mercury News reporting on the upcoming RoboNexus conference. Like other robot articles in the Silicon Valley press, the rise of the robots at the weekend trade show is treated like a report on a strange circus carnival. The surprising fact that robots exist is reported, coupled with happy re-assurances that nothing will happen for many years and we're a long way away from anything a Silicon Valley person would care about.
Why the soothing message of robotic irrelevance? If robots are really on the rise, Silicon Valley is in trouble. Clearly, the average Silicon Valley computer developer has no more knowledge of robotics than anyone else - bad news for this group ever "leading" the rise of the robots.
A few cool companies like Anybots (humanoids) and Team Overbot (DARPA Grand Challenge for robot-cars) are working in the Silicon Valley area. But they are islands in a sea of groups working on "more important" things (presumably another try at ITV and Linux) that are linear continuations of the last 20 years of computers and cyberspace. These dorks are slavering over Bluetooth in your pen or the latest crazy hiring questionnaire from Google (uncool, non-Matrix-lovin', non-digiteratti who don't read Wired magazine need not apply).
Of course, the detractors of robotics have a point. When the PC first appeared in the early 1980s, some assumed that the HERO-ic age of robotics had dawned, and hobbyists began creating simple robots while professionals put embedded microcontrollers into cars and microwaves. They're still doing this - often with the same 20-year old, 4MHz 1Kb memory microcontrollers they used then. The fun was all in the creation and understanding embedded systems - hardly a mass-market preoccupation. During the 1980s and 1990s, both research robots and hobby robots were the source of periodic press releases concerning slightly laughable machines with no connection to anything the non-specialist would care about. Meanwhile, Hollywood told us that "robots" were formerly unemployed SAG actors with grey makeup and a mean face.
Today, we're seeing a different world. The U.N. Economic Commission for Europe and the International Federation of Robotics just reported reported 607,000 automated domestic helpers were in use at the end of 2003, 67% of them purchased that year. Most of them - 570,000 - were robot lawnmowers. Sales of vacuum cleaning robots were reported as 37,000, (but this is false - iRobot has sold over 500,000 Roombas). According to the report, by 2008, some 4.1 mln domestic robots will be in use, the study said. Service, entertainment, and domestic robots will soon eclipse industrial robots, currently at about 1 million worldwide.
7-fold in 4 years...that's full-bore growth like the PC industry experienced in the 1980s, and much faster than the tepid growth of e-commerce (up about 20% in 2004) or Internet usage (it actually dropped during the first half of 2004).
And where is the investment? The Mercury article notes that VCs are "uninterested" in robotics at present - I suppose they're still waiting for the next company to try sending "movies over your cellphone" or another useless thing
...those were the days, my friend, we won't admit that they reached the end...(music)
I suspect that "faster" or "beter" hardware is not the reason for this spurt in robots. The iRobot Roomba uses the same components that a hobby robot does, but it does something useful. It is loaded with top-flight engineering and industrial design aimed at consumers. There's been a change in mindset about robotics, allowing useful products to appear.
Possibly, with the waning of cyberspace as a "hot thing" and the decidedly un-newsworthy spread of recent "big next things" such as wireless, there is at last space in the collective unconscious for a different computing paradigm. One thing that is clear - robotics is not just another kind of PC, even though robotics companies are stealing PC technology for their decidedly un-PC creations. Programmers of the virtual world will have a hard time making the jump - but a few will. With the success of a few platforms (e.g., Friendly Robotics lawnmowers, and home-bots in Korea and Japan, entertainment "super-toys" like the Sony Aibo) the industry is now waking up and realizing that the goal of building a robot is not to use a single 1980s microcontroller for a proud display of line-following.
PC robots using a standard personal computer OS like White Box Robotics have been attacked for the use of a sloppy, wasteful operating system like Windows. Who cares, except an embedded systems purist? If it works, use it - hijack it for your own purposes. The early personal computer industry hijacked microprocessors intended for other purposes. The PC-derived robot is one of several attempts to develop standard hardware and software platforms for consumer robotics and put a standardized "floor" under at least one class of consumer-bot. The point is to get something out there that works, that non-embedded programmers can use, without constantly re-inventing the wheel. As the industry grows, it will be possible for hardware developers to make chips and other hardware specifically for robots, and we'll see the real explosion.
I suspect that robotics will soon develop its own OS family (not based on standard embedded OSes, either) but for now, creating first-generations Robots That Move is the priority. Hardware that can do more than do more than crawl across a lab floor in 4 hours is a priority. Cheap legs from Anybots and/or balancing-wheel bots like the Segway RMP platform are what's needed for Silicon Valley to finally wake up and smell the robot.
Wednesday, October 20, 2004
Segway Robot Rising
Interesting article today over at space.com, concerning the modification of the original Segway "people mover" into a dynamically-balanced base for wheeled robots. I know this blog is about Robots that Jump, but using a Segway RMP platform for wheeled motion is very different than the standard "engineering school" wheeled robot.
For one thing, the wheels are big, and designed for outdoor use. This allows a Robotic Mobility Platform (RMP)-based robot to handle small cracks in the sidewalk, grass, and other environments that would be impossible for your typical lab-bot. The RMP is designed for consumer use, meaning that it is rugged and unlikely to break down in challenging environments. In contrast, many wheeled lab-bots spend most of their life "broken."
Second, the system is dynamically balanced, which makes it move fast, efficiently, and in a lifelike manner. To move, a RMP robot will send commands causing it to tilt forward or back, resulting in a smooth speed change. Having only two wheels, it takes up less surface area, and can potentially navigate through narrow halls and doors. Top speed is very impressive 8mph, compared to lab-bots inch a minute.
Third, the system can carry up to 100 pounds of equipment. This allows for a huge amount of computer compared to a typical lab-bot or hobby robot. There's room for several high-performance PC boards plus batteries and sensors with weight to spare.
Fourth, the sysetm is already battery-operated, and capable of significant range.
The RMP has the potential to be a "breakthrough" robotic technology because it also imposes something very important: standardization of components. Currently, most robots are completely custom-built, one-of-a-kind machines. This makes it difficult to implement advances in hardware and software on one machine onto another. Without standardization, robots are doomed to be grad student pets or hobbytrons and will never develop the strength or reliability needed for consumer use. One can easily imagine an RMP moving quickly in large buildings (old Segway movies showed people using the machines in rooms), lawns, back lots, even gardens and greenhouses.
There's even the potential for smaller, two-wheeled robots using Segway-type technology. Already, in Korea, a company called Mostitech at http://www.mostitech.com is selling a much smaller two-wheeled robot. One can imagine this company creating the base platform, and selling it to other robotics companies to develop their own "upper bodies."
Toyota has also developed a two-wheeled robot. The lower body is much like a Segway, but the upper body is similar to its two-legged robot recently demonstrated playing a trumpet (really blowing air and working the trumpet valves, by the way).
The availability of the RMP puts robotics developers on the spot. For decades, limited mobility has been something that developers could always point to when explaining a robot's failure to do something useful. Now, anyone can get an RMP and have a automated, self-balancing, agile transportation system out of the box. The challenge now shjifts to making the robot do something useful - no need to re-invent the wheel all over again.
According to one person quoted in the article:
"We have the motion problem solved," Veloso told SPACE.com. "So we canfocus on the other aspects."
There's no excuse left for not fixing those "other aspects."
Interesting article today over at space.com, concerning the modification of the original Segway "people mover" into a dynamically-balanced base for wheeled robots. I know this blog is about Robots that Jump, but using a Segway RMP platform for wheeled motion is very different than the standard "engineering school" wheeled robot.
For one thing, the wheels are big, and designed for outdoor use. This allows a Robotic Mobility Platform (RMP)-based robot to handle small cracks in the sidewalk, grass, and other environments that would be impossible for your typical lab-bot. The RMP is designed for consumer use, meaning that it is rugged and unlikely to break down in challenging environments. In contrast, many wheeled lab-bots spend most of their life "broken."
Second, the system is dynamically balanced, which makes it move fast, efficiently, and in a lifelike manner. To move, a RMP robot will send commands causing it to tilt forward or back, resulting in a smooth speed change. Having only two wheels, it takes up less surface area, and can potentially navigate through narrow halls and doors. Top speed is very impressive 8mph, compared to lab-bots inch a minute.
Third, the system can carry up to 100 pounds of equipment. This allows for a huge amount of computer compared to a typical lab-bot or hobby robot. There's room for several high-performance PC boards plus batteries and sensors with weight to spare.
Fourth, the sysetm is already battery-operated, and capable of significant range.
The RMP has the potential to be a "breakthrough" robotic technology because it also imposes something very important: standardization of components. Currently, most robots are completely custom-built, one-of-a-kind machines. This makes it difficult to implement advances in hardware and software on one machine onto another. Without standardization, robots are doomed to be grad student pets or hobbytrons and will never develop the strength or reliability needed for consumer use. One can easily imagine an RMP moving quickly in large buildings (old Segway movies showed people using the machines in rooms), lawns, back lots, even gardens and greenhouses.
There's even the potential for smaller, two-wheeled robots using Segway-type technology. Already, in Korea, a company called Mostitech at http://www.mostitech.com is selling a much smaller two-wheeled robot. One can imagine this company creating the base platform, and selling it to other robotics companies to develop their own "upper bodies."
Toyota has also developed a two-wheeled robot. The lower body is much like a Segway, but the upper body is similar to its two-legged robot recently demonstrated playing a trumpet (really blowing air and working the trumpet valves, by the way).
The availability of the RMP puts robotics developers on the spot. For decades, limited mobility has been something that developers could always point to when explaining a robot's failure to do something useful. Now, anyone can get an RMP and have a automated, self-balancing, agile transportation system out of the box. The challenge now shjifts to making the robot do something useful - no need to re-invent the wheel all over again.
According to one person quoted in the article:
"We have the motion problem solved," Veloso told SPACE.com. "So we canfocus on the other aspects."
There's no excuse left for not fixing those "other aspects."
