For dexterity, manipulative skill and and sheer engineering complexity, it’s hard to go past this high-speed robot hand.
3 comments:
Falafulu Fisi
said...
That's a learning robot , where it adapts its action on the fly very very fast. There is no pre-programmed procedures that is embedded in its system, since there are different millions possible scenario that one could write lines of codes to anticipate of how that tennis ball may fall or bounce in 3D.
Try asking a programmer if he could write some traditional (ie, non-learning) software to imitate what that robot does in real time, and I bet you that he/she would say, that's undoable. This task is doable if the software is written to learn and adapt.
For those who hung up on the notion that the robot learns should not worry, because their leaning task is still primitive that they won't dominate us in the near future, but actually, the way they're being programmed is actually the same way humans learn.
If anyone doubts this, then try defining the term learning in a quantifiable manner, because the matter of fact is, the learning process that occurs in the brain is in fact quantifiable, but its just that we use words semantic (cognitively / philosophical) to describe them, not mathematical ways.
So, no one should try to be hung up too much on the concept that machine can learn, because once anyone can define learning (as in humans, etc...), he would be awestruck to find that his definition fits in with what that robot is doing, without him/her knowing it.
Human learning is obviously more complex, that arised from smaller tasks that build up to form a complex learning process for a specific task (eg, learning calculus). What that robot is doing is just one single task, where as humans we're multi-task. If one compares the single learning task that is part of a complex learning phenomena for humans such as calculus, he/she would find that the single task learning that guides the robot's action is the same as that single task learning that contributes to many , if not thousands of other single learning processes that comprise that complex calculus understanding/learning.
For dexterity, manipulative skill and and sheer engineering complexity, it’s hard to go past this high-speed robot hand.
Yes, it's amazing the direction robotics is going. With all that dribbling, catching and throwing it struck me that one day soon there will be robotic sports leagues (perhaps an NRBA instead of the NBA for example). Sure these machines are fast, but still, only a tiny fraction of a second faster than this guy:
3 comments:
That's a learning robot , where it adapts its action on the fly very very fast. There is no pre-programmed procedures that is embedded in its system, since there are different millions possible scenario that one could write lines of codes to anticipate of how that tennis ball may fall or bounce in 3D.
Try asking a programmer if he could write some traditional (ie, non-learning) software to imitate what that robot does in real time, and I bet you that he/she would say, that's undoable. This task is doable if the software is written to learn and adapt.
For those who hung up on the notion that the robot learns should not worry, because their leaning task is still primitive that they won't dominate us in the near future, but actually, the way they're being programmed is actually the same way humans learn.
If anyone doubts this, then try defining the term learning in a quantifiable manner, because the matter of fact is, the learning process that occurs in the brain is in fact quantifiable, but its just that we use words semantic (cognitively / philosophical) to describe them, not mathematical ways.
So, no one should try to be hung up too much on the concept that machine can learn, because once anyone can define learning (as in humans, etc...), he would be awestruck to find that his definition fits in with what that robot is doing, without him/her knowing it.
Human learning is obviously more complex, that arised from smaller tasks that build up to form a complex learning process for a specific task (eg, learning calculus). What that robot is doing is just one single task, where as humans we're multi-task. If one compares the single learning task that is part of a complex learning phenomena for humans such as calculus, he/she would find that the single task learning that guides the robot's action is the same as that single task learning that contributes to many , if not thousands of other single learning processes that comprise that complex calculus understanding/learning.
By definition, the robot is a learning robot.
Can this be taught to pick up the toys before the iRobot vacuums?
For dexterity, manipulative skill and and sheer engineering complexity, it’s hard to go past this high-speed robot hand.
Yes, it's amazing the direction robotics is going. With all that dribbling, catching and throwing it struck me that one day soon there will be robotic sports leagues (perhaps an NRBA instead of the NBA for example). Sure these machines are fast, but still, only a tiny fraction of a second faster than this guy:
http://www.youtube.com/watch?v=wLnmvseCseI
:^)
Post a Comment