Saturday, October 09, 2004
I'll be at RoboNexus, farm-bots, 3D printing, QRIO, and robots in public spaces
I am going to be at RoboNexus the third week of October on Friday (session) and Saturday (Expo). Hope - to see you there! I'm particularly interested in contacting people about moving robots into the education of design students such as Industrial Design and Web/interactive device design. As the robotics industry grows there will be a need for industrial and web/interactive designers who understand robots.
A few interesting articles on Robotic Trends. First, one on automated farms. Apparently, GPS-guided tractors are a coming thing. One advantage of a robotic tractor: it can be much smaller and more efficient, since it doesn't have to have a cab for a human operator. Researchers at the University of Illinois are taking the hobby robot/junkbot approach - creating small swarms of agricultural robots with available parts just lying around. Other researchers are also cited in Germany and Japan - all building robots that cost a few hundred to thousand dollars.
In other words, we've found a reasonable use for hobby robots - farming! This has always been puzzling to me - we have cheap hobby robots built under a 'junkbot' paradigm, and ultra-expensive research robots, with no middle ground. A middle ground will be essential for the industry to grow. The question is whether the research models become cheaper, or the hobby robots become more capable. At present, the second seems to be true.
This brings up an important point about Robots That Jump. One of the key problems of the industry today is that robot designers are generally restricted to pre-manufactured body components. That is, it is difficult to build a robot body with an arbitrary design, since the parts have to be custom-made. I am sure that part of the huge cost of developing the Sony QURO and Honda Asimo was making custom parts for these machines.
Fortunately, we seem to be on the verge of 'fixing' this problem. With the advent of 3D printers like those from Dimension Printing (cheap), Zcorp (high-end, nice flash animation of printer here), it should be possible to prototype a robot body directly from a CAD/CAM engine. The newest 3D printers allow moving parts to be formed during the printing process, and are less restricted in the kinds of materials used to create the 3D print. In addition, new, fully 3D CAD/CAM allows preliminary "physics" testing of designed objects.
One can imagine a future robotic 'birthing' workstation (like the one featured at the 2003 Robodex with Astroboy). The robot body is designed and virtual physics tested. Then, a mix of 3D printing plus robotic arm assembly creates the robot. It wouldn't necessarily be complete parts. One can imagine the 3D printer imaging head making a few passes to create a hollow area in the part. A robot arm would locate the hollow and insert another part. The printer would continue, embedding the part. This approach would be ideal for integrating large numbers of sensors into the robotic shell. Another possiblity would be for the robot arm to pick up a part, rotate it, snap/screw in components, and then place it back on a 3D printer for further work. This opens up novel ideas for robot component design.
Software would need to be developed that can break down the designed body into the components which should be 3D-printed, and those which can be welded/bolted together, but this seems to be a semi-automatic process. Some research at Brandeis university a few years ago used this approach - a genetic algorithm 'designed' robot bodies, some of which were automatically created in 3D material. The system operated the resulting robot bodies and used the feedback to re-design if necessary.
Discussing this concept makes me think of some recent information I discovered about the Sony QURO. Any Robot That Jumps will have to have a body awareness - it will need to sense the status of its body, and be able to take protective action if it might be damaged. It will also need to be aware of getting dirty, and inspect/clean itself regularly. It turns out that the QRIO does have some of these qualities. When the robot falls over, it tries to land in a way that protects its head, where delicate cameras are present. Interesting!
Finally, a comment about robots in public spaces. One of the problems with robots in our current society is that there is no legal precedent for autonomous robots operating in public areas. The DARPA Grand Challenge in 2004 had numerous restrictions, most of which were related to not letting the robo-cars venture into desert wilderness. Similar problems will emerge when we begin testing robots in malls, public buildings, and on the street. According to an article on Robotics Trends, Japan is addressing the issue by creating toku - public de-regulation zones.
To quote:
"...Believing the situation should be changed, Takamoto got involved in the prefectural government's efforts to obtain robot tokku authorization. In the first project to benefit from the special deregulation, tmsuk's remote-controlled robot tmsuk04 "went shopping" during a field test on the busy Kami-Kawabata Shotengai shopping street in Hakata Ward, Fukuoka, in February. In the test, the robot, which can be operated over a public mobile phone network with a data transmission speed of 64 kilobytes per second--a service widely available in Japan--picked up a bag from a wagon outside a shop..."
I suspect such a test would be next to impossible in the US at present. This is one of many things hindering US adoption of Robots That Jump.
I am going to be at RoboNexus the third week of October on Friday (session) and Saturday (Expo). Hope - to see you there! I'm particularly interested in contacting people about moving robots into the education of design students such as Industrial Design and Web/interactive device design. As the robotics industry grows there will be a need for industrial and web/interactive designers who understand robots.
A few interesting articles on Robotic Trends. First, one on automated farms. Apparently, GPS-guided tractors are a coming thing. One advantage of a robotic tractor: it can be much smaller and more efficient, since it doesn't have to have a cab for a human operator. Researchers at the University of Illinois are taking the hobby robot/junkbot approach - creating small swarms of agricultural robots with available parts just lying around. Other researchers are also cited in Germany and Japan - all building robots that cost a few hundred to thousand dollars.
In other words, we've found a reasonable use for hobby robots - farming! This has always been puzzling to me - we have cheap hobby robots built under a 'junkbot' paradigm, and ultra-expensive research robots, with no middle ground. A middle ground will be essential for the industry to grow. The question is whether the research models become cheaper, or the hobby robots become more capable. At present, the second seems to be true.
This brings up an important point about Robots That Jump. One of the key problems of the industry today is that robot designers are generally restricted to pre-manufactured body components. That is, it is difficult to build a robot body with an arbitrary design, since the parts have to be custom-made. I am sure that part of the huge cost of developing the Sony QURO and Honda Asimo was making custom parts for these machines.
Fortunately, we seem to be on the verge of 'fixing' this problem. With the advent of 3D printers like those from Dimension Printing (cheap), Zcorp (high-end, nice flash animation of printer here), it should be possible to prototype a robot body directly from a CAD/CAM engine. The newest 3D printers allow moving parts to be formed during the printing process, and are less restricted in the kinds of materials used to create the 3D print. In addition, new, fully 3D CAD/CAM allows preliminary "physics" testing of designed objects.
One can imagine a future robotic 'birthing' workstation (like the one featured at the 2003 Robodex with Astroboy). The robot body is designed and virtual physics tested. Then, a mix of 3D printing plus robotic arm assembly creates the robot. It wouldn't necessarily be complete parts. One can imagine the 3D printer imaging head making a few passes to create a hollow area in the part. A robot arm would locate the hollow and insert another part. The printer would continue, embedding the part. This approach would be ideal for integrating large numbers of sensors into the robotic shell. Another possiblity would be for the robot arm to pick up a part, rotate it, snap/screw in components, and then place it back on a 3D printer for further work. This opens up novel ideas for robot component design.
Software would need to be developed that can break down the designed body into the components which should be 3D-printed, and those which can be welded/bolted together, but this seems to be a semi-automatic process. Some research at Brandeis university a few years ago used this approach - a genetic algorithm 'designed' robot bodies, some of which were automatically created in 3D material. The system operated the resulting robot bodies and used the feedback to re-design if necessary.
Discussing this concept makes me think of some recent information I discovered about the Sony QURO. Any Robot That Jumps will have to have a body awareness - it will need to sense the status of its body, and be able to take protective action if it might be damaged. It will also need to be aware of getting dirty, and inspect/clean itself regularly. It turns out that the QRIO does have some of these qualities. When the robot falls over, it tries to land in a way that protects its head, where delicate cameras are present. Interesting!
Finally, a comment about robots in public spaces. One of the problems with robots in our current society is that there is no legal precedent for autonomous robots operating in public areas. The DARPA Grand Challenge in 2004 had numerous restrictions, most of which were related to not letting the robo-cars venture into desert wilderness. Similar problems will emerge when we begin testing robots in malls, public buildings, and on the street. According to an article on Robotics Trends, Japan is addressing the issue by creating toku - public de-regulation zones.
To quote:
"...Believing the situation should be changed, Takamoto got involved in the prefectural government's efforts to obtain robot tokku authorization. In the first project to benefit from the special deregulation, tmsuk's remote-controlled robot tmsuk04 "went shopping" during a field test on the busy Kami-Kawabata Shotengai shopping street in Hakata Ward, Fukuoka, in February. In the test, the robot, which can be operated over a public mobile phone network with a data transmission speed of 64 kilobytes per second--a service widely available in Japan--picked up a bag from a wagon outside a shop..."
I suspect such a test would be next to impossible in the US at present. This is one of many things hindering US adoption of Robots That Jump.
I'll be at RoboNexus, farm-bots, 3D printing, QRIO, and robots in public spaces
I am going to be at RoboNexus the third week of October on Friday (session) and Saturday (Expo). Hope - to see you there! I'm particularly interested in contacting people about moving robots into the education of design students such as Industrial Design and Web/interactive device design. As the robotics industry grows there will be a need for industrial and web/interactive designers who understand robots.
A few interesting articles on Robotic Trends. First, one on automated farms. Apparently, GPS-guided tractors are a coming thing. One advantage of a robotic tractor: it can be much smaller and more efficient, since it doesn't have to have a cab for a human operator. Researchers at the University of Illinois are taking the hobby robot/junkbot approach - creating small swarms of agricultural robots with available parts just lying around. Other researchers are also cited in Germany and Japan - all building robots that cost a few hundred to thousand dollars.
In other words, we've found a reasonable use for hobby robots - farming! This has always been puzzling to me - we have cheap hobby robots built under a 'junkbot' paradigm, and ultra-expensive research robots, with no middle ground. A middle ground will be essential for the industry to grow. The question is whether the research models become cheaper, or the hobby robots become more capable. At present, the second seems to be true.
This brings up an important point about Robots That Jump. One of the key problems of the industry today is that robot designers are generally restricted to pre-manufactured body components. That is, it is difficult to build a robot body with an arbitrary design, since the parts have to be custom-made. I am sure that part of the huge cost of developing the Sony QURO and Honda Asimo was making custom parts for these machines.
Fortunately, we seem to be on the verge of 'fixing' this problem. With the advent of 3D printers like those from Dimension Printing (cheap), Zcorp (high-end, nice flash animation of printer here), it should be possible to prototype a robot body directly from a CAD/CAM engine. The newest 3D printers allow moving parts to be formed during the printing process, and are less restricted in the kinds of materials used to create the 3D print. In addition, new, fully 3D CAD/CAM allows preliminary "physics" testing of designed objects.
One can imagine a future robotic 'birthing' workstation (like the one featured at the 2003 Robodex with Astroboy). The robot body is designed and virtual physics tested. Then, a mix of 3D printing plus robotic arm assembly creates the robot. It wouldn't necessarily be complete parts. One can imagine the 3D printer imaging head making a few passes to create a hollow area in the part. A robot arm would locate the hollow and insert another part. The printer would continue, embedding the part. This approach would be ideal for integrating large numbers of sensors into the robotic shell. Another possiblity would be for the robot arm to pick up a part, rotate it, snap/screw in components, and then place it back on a 3D printer for further work. This opens up novel ideas for robot component design.
Software would need to be developed that can break down the designed body into the components which should be 3D-printed, and those which can be welded/bolted together, but this seems to be a semi-automatic process. Some research at Brandeis university a few years ago used this approach - a genetic algorithm 'designed' robot bodies, some of which were automatically created in 3D material. The system operated the resulting robot bodies and used the feedback to re-design if necessary.
Discussing this concept makes me think of some recent information I discovered about the Sony QURO. Any Robot That Jumps will have to have a body awareness - it will need to sense the status of its body, and be able to take protective action if it might be damaged. It will also need to be aware of getting dirty, and inspect/clean itself regularly. It turns out that the QRIO does have some of these qualities. When the robot falls over, it tries to land in a way that protects its head, where delicate cameras are present. Interesting!
Finally, a comment about robots in public spaces. One of the problems with robots in our current society is that there is no legal precedent for autonomous robots operating in public areas. The DARPA Grand Challenge in 2004 had numerous restrictions, most of which were related to not letting the robo-cars venture into desert wilderness. Similar problems will emerge when we begin testing robots in malls, public buildings, and on the street. According to an article on Robotics Trends, Japan is addressing the issue by creating toku - public de-regulation zones.
To quote:
"...Believing the situation should be changed, Takamoto got involved in the prefectural government's efforts to obtain robot tokku authorization. In the first project to benefit from the special deregulation, tmsuk's remote-controlled robot tmsuk04 "went shopping" during a field test on the busy Kami-Kawabata Shotengai shopping street in Hakata Ward, Fukuoka, in February. In the test, the robot, which can be operated over a public mobile phone network with a data transmission speed of 64 kilobytes per second--a service widely available in Japan--picked up a bag from a wagon outside a shop..."
I suspect such a test would be next to impossible in the US at present. This is one of many things hindering US adoption of Robots That Jump.
I am going to be at RoboNexus the third week of October on Friday (session) and Saturday (Expo). Hope - to see you there! I'm particularly interested in contacting people about moving robots into the education of design students such as Industrial Design and Web/interactive device design. As the robotics industry grows there will be a need for industrial and web/interactive designers who understand robots.
A few interesting articles on Robotic Trends. First, one on automated farms. Apparently, GPS-guided tractors are a coming thing. One advantage of a robotic tractor: it can be much smaller and more efficient, since it doesn't have to have a cab for a human operator. Researchers at the University of Illinois are taking the hobby robot/junkbot approach - creating small swarms of agricultural robots with available parts just lying around. Other researchers are also cited in Germany and Japan - all building robots that cost a few hundred to thousand dollars.
In other words, we've found a reasonable use for hobby robots - farming! This has always been puzzling to me - we have cheap hobby robots built under a 'junkbot' paradigm, and ultra-expensive research robots, with no middle ground. A middle ground will be essential for the industry to grow. The question is whether the research models become cheaper, or the hobby robots become more capable. At present, the second seems to be true.
This brings up an important point about Robots That Jump. One of the key problems of the industry today is that robot designers are generally restricted to pre-manufactured body components. That is, it is difficult to build a robot body with an arbitrary design, since the parts have to be custom-made. I am sure that part of the huge cost of developing the Sony QURO and Honda Asimo was making custom parts for these machines.
Fortunately, we seem to be on the verge of 'fixing' this problem. With the advent of 3D printers like those from Dimension Printing (cheap), Zcorp (high-end, nice flash animation of printer here), it should be possible to prototype a robot body directly from a CAD/CAM engine. The newest 3D printers allow moving parts to be formed during the printing process, and are less restricted in the kinds of materials used to create the 3D print. In addition, new, fully 3D CAD/CAM allows preliminary "physics" testing of designed objects.
One can imagine a future robotic 'birthing' workstation (like the one featured at the 2003 Robodex with Astroboy). The robot body is designed and virtual physics tested. Then, a mix of 3D printing plus robotic arm assembly creates the robot. It wouldn't necessarily be complete parts. One can imagine the 3D printer imaging head making a few passes to create a hollow area in the part. A robot arm would locate the hollow and insert another part. The printer would continue, embedding the part. This approach would be ideal for integrating large numbers of sensors into the robotic shell. Another possiblity would be for the robot arm to pick up a part, rotate it, snap/screw in components, and then place it back on a 3D printer for further work. This opens up novel ideas for robot component design.
Software would need to be developed that can break down the designed body into the components which should be 3D-printed, and those which can be welded/bolted together, but this seems to be a semi-automatic process. Some research at Brandeis university a few years ago used this approach - a genetic algorithm 'designed' robot bodies, some of which were automatically created in 3D material. The system operated the resulting robot bodies and used the feedback to re-design if necessary.
Discussing this concept makes me think of some recent information I discovered about the Sony QURO. Any Robot That Jumps will have to have a body awareness - it will need to sense the status of its body, and be able to take protective action if it might be damaged. It will also need to be aware of getting dirty, and inspect/clean itself regularly. It turns out that the QRIO does have some of these qualities. When the robot falls over, it tries to land in a way that protects its head, where delicate cameras are present. Interesting!
Finally, a comment about robots in public spaces. One of the problems with robots in our current society is that there is no legal precedent for autonomous robots operating in public areas. The DARPA Grand Challenge in 2004 had numerous restrictions, most of which were related to not letting the robo-cars venture into desert wilderness. Similar problems will emerge when we begin testing robots in malls, public buildings, and on the street. According to an article on Robotics Trends, Japan is addressing the issue by creating toku - public de-regulation zones.
To quote:
"...Believing the situation should be changed, Takamoto got involved in the prefectural government's efforts to obtain robot tokku authorization. In the first project to benefit from the special deregulation, tmsuk's remote-controlled robot tmsuk04 "went shopping" during a field test on the busy Kami-Kawabata Shotengai shopping street in Hakata Ward, Fukuoka, in February. In the test, the robot, which can be operated over a public mobile phone network with a data transmission speed of 64 kilobytes per second--a service widely available in Japan--picked up a bag from a wagon outside a shop..."
I suspect such a test would be next to impossible in the US at present. This is one of many things hindering US adoption of Robots That Jump.
