Robots problems and their mapping and vision are all topics on this episode of The AI Minute. For more on Artificial Intelligence: https://voicesinai.com https://gigaom.com https://byronreese.com https://amzn.to/2vgENbn... Transcript: Robots haven’t developed anywhere near as quickly as computers. There are so many challenges involved in building them. Let’s look at just two of them. The first challenge for robots is perceiving their environment. They do this, of course, through sensors. Humans have a pretty robust set of sensors ourselves, but generally speaking, robots can measure more things with more precision than humans. But that doesn’t actually mean they are better at perceiving things than we are. Take, for instance, the problem of vision. This is an incredibly hard problem for robots. The number of things going on in your brain when you glance down a hallway is complex in the extreme. A description of how your brain performs that minor miracle would require pages and pages of technobabble about polygons and cones and layers and all the rest. So even though you can clamp a high definition HD camera on to a robot, that doesn’t give it a high degree of perception, rather, it just provides a whole lot of data that the robot still has to make sense of. Robots also have trouble figuring out where they are. Roboticists don’t really have best practices around how to do this. It really varies by the situation. Often a robot is tasked with making a map of where it is, then keeping track of where it is on that map. This probably doesn’t sound too hard, because we do it effortlessly. But imagine the problem from the point of view of a robot. A robot has been dropped into a room. It sees a chair and a footstool. Since the chair and the footstool can be moved, the robot can’t use them as anchors. It is in a constant existential crisis of “Did I move? Or did the chair move?†As such, it constantly has to redraw its map. Building a map and figuring out where you are on that map is called SLAM (simultaneous localization and mapping). While it isn’t an insurmountable problem by a long shot, it’s just another thing that makes the job of being roboticist a challenging one.
2019年5月23日星期四
Robot vs Computers: Who Wins the Race?
Robots problems and their mapping and vision are all topics on this episode of The AI Minute. For more on Artificial Intelligence: https://voicesinai.com https://gigaom.com https://byronreese.com https://amzn.to/2vgENbn... Transcript: Robots haven’t developed anywhere near as quickly as computers. There are so many challenges involved in building them. Let’s look at just two of them. The first challenge for robots is perceiving their environment. They do this, of course, through sensors. Humans have a pretty robust set of sensors ourselves, but generally speaking, robots can measure more things with more precision than humans. But that doesn’t actually mean they are better at perceiving things than we are. Take, for instance, the problem of vision. This is an incredibly hard problem for robots. The number of things going on in your brain when you glance down a hallway is complex in the extreme. A description of how your brain performs that minor miracle would require pages and pages of technobabble about polygons and cones and layers and all the rest. So even though you can clamp a high definition HD camera on to a robot, that doesn’t give it a high degree of perception, rather, it just provides a whole lot of data that the robot still has to make sense of. Robots also have trouble figuring out where they are. Roboticists don’t really have best practices around how to do this. It really varies by the situation. Often a robot is tasked with making a map of where it is, then keeping track of where it is on that map. This probably doesn’t sound too hard, because we do it effortlessly. But imagine the problem from the point of view of a robot. A robot has been dropped into a room. It sees a chair and a footstool. Since the chair and the footstool can be moved, the robot can’t use them as anchors. It is in a constant existential crisis of “Did I move? Or did the chair move?†As such, it constantly has to redraw its map. Building a map and figuring out where you are on that map is called SLAM (simultaneous localization and mapping). While it isn’t an insurmountable problem by a long shot, it’s just another thing that makes the job of being roboticist a challenging one.
订阅：
博文评论 (Atom)
没有评论:
发表评论