Perceiving Robotic Industrial Arm Inches Toward Autonomy 22
kkleiner writes "Willow Garage spinoff IPI has developed a visual system for its line of robotic arms that enable the machines to perceive a specific object in the midst of random ones. On-site videos show the 'sensing' robots analyzing stacks of random boxes, selecting certain ones, and tossing them to a human handler. The software is also used in an automated box unloader that requires no human supervision."
"Perceiving Robotic Industrial Arm" (Score:2)
Luxo Jr. (Score:3)
That second video is just great. The robot really looks like it's an animal trying to figure out how to get at the box.
It makes me wonder if we can get some Pixar animators to work on these robots when it comes to human interaction. I'd love to see a robot arm look all dejected when it can't figure out the problem, and then sheepishly ask its meatbag handler for help.
Re:Luxo Jr. (Score:2)
I kept waiting for the human catching the boxes to explode. Maybe someone could whip up a quick and dirty edit.
Seems little different than 1990s tech ... (Score:2)
Finding parts at any orientation amongst noise, parts that may be partially obscured. That was also on the factory floors in the 90s.
Stacking and unstacking boxes/items to get to a desired box/item, again, 90s. Tossing the box to a person, perhaps that is new.
Re:Seems little different than 1990s tech ... (Score:2)
True enough in the lab. This is industrial robotics, where reliability is key.
Re:Seems little different than 1990s tech ... (Score:2)
True enough in the lab. This is industrial robotics, where reliability is key.
Finding and picking parts among noise and clutter, parts at any orientation, parts that may be partially obscured by other parts laying on top of them. That was on factory floors in the 1990s.
Re:Seems little different than 1990s tech ... (Score:0)
yeah, but could the robot find it and pick it up?
Re:Seems little different than 1990s tech ... (Score:2)
yeah, but could the robot find it and pick it up?
Yes. For example in one application they were picking parts for assembly. Changing the orientation of the parts as necessary and placing it into a jig.
Take note ups, fed-ex, usps workers. (Score:2)
This will soon eliminate your jobs.
Re:Take note ups, fed-ex, usps workers. (Score:3, Funny)
Re:Take note ups, fed-ex, usps workers. (Score:1)
Relevant Textbook (Score:2)
01010101 01101110 01100011 01101100 01100101 01000 (Score:-1)
Try less whitespace and/or less repetition
01000111 01101001 01110110 01100101 0100000 01101101 01100101 0100000 01111001 01101111 01110101 01110010 0100000 01101100 01110101 01101110 01100011 01101000 0100000 01101101 01101111 01101110 01100101 01111001 0101100 0100000 01110000 01110101 01101110 01101011 0100001
Moving boxes is okay, I guess (Score:2)
I would prefer something more useful [wordpress.com]
wonder if this can be used for sorting recycling (Score:2)
One of the problems with "single-stream" waste management, which otherwise has a number of advantages, is the difficulty of automatically sorting out the recyclables from what results. Metals are relatively easy to pick out, using either magnetism (for ferrous metals) or induced magnetism via eddy currents (mainly for aluminum), but a good deal of sorting still has to be done in a somewhat manual fashion, which is difficult for humans to do, partly because there might be dangerous stuff amidst the waste stream. A robot using computer vision to do some additional sorting beyond what can be done using characteristics like magnetism and weight could be a nice addition.
Re:wonder if this can be used for sorting recyclin (Score:2)
There is work being done by a bunch of people in the EU, mainly due to the great legislation regarding waste reduction. There's a great little book I picked up a bit ago outlining some of the processes in automated waste recovery [slashdot.org] (Comprehensive Information Chain for Automated Disassembly of Obsolete Technical Appliances ) . I had gone into grad school to work on this area, and it turns out it's been going on for the past 30 years or more.
The biggest issue is cost, where it's still cheaper for manual labor than to automate the system. If you're interested in some papers I have a whole boatload regarding automation in electronic waste.
Is there an autodidact in the house? (Score:1)
Re:Is there an autodidact in the house? (Score:0)
And what motivations will a robot capable of learning on its own have? Why is the default assumption always that intelligent robots will have ape-like motivations and seek to be alpha or kill off the other tribe?
Perceiving is easy... (Score:3)
Integrating what is perceived is the HARD PROBLEM.
Sorry -- presently watching, in order, the MIT150 'Brains,
Minds and Intelligence' video-series.
(I can really recommend the series -- at least it taught
me how still, a.d. 2013, much of AI is still a lot of PR).
Enjoy!
Robot Finds Kitten (Score:0)
... that would be the best demo for this :)
Re:Robot Finds Kitten (Score:3)
I think you mean 'robot flings kitten'.
More jobs lost (Score:0)
Once this is perfected, even more jobs will be lost to "automation" and the bottom line.