because the modern world is built around people shaped like people. all the tools and workspaces and interfaces already optimized for it. and that keeps it safe for prople to co-exist. if we start building the world around some otherness, then humans are locked out and obsolete.
right, but not every place we’re going to use them is a factory. And it’s possible we’re going to use them in places we’d still like to be human-first because of their critical role, such as i.e. dams
That’s just not true, you’d have to completely rebuild your entire production facilities which would cost more in the long run than taking say… Boston Dynamics Atlas and hooking it up to an LLM trained on a specific task set.
Newer facilities could be built for the future where humans aren’t involved at all, but in the interim making robots that move and manipulate objects like we do is still the better solution.
It is very difficult to make a robotic hand that can operate a screwdriver. If the robot only ever needs to perform one task on an assembly line, just build it with the tools as part of it. Of course, some modularity helps to retool the plant for another product but there are very few cases where a robot needs the versatility of the human hand (maybe bomb defusal?) or body.
yes we have millions of tethered robotic arms in use today with tools as hands. but once an untethered humanoid robot is available on the market with robotic hands, the use cases are infinite.
I don’t think it’s productive to try replicating the human hand accurately enough to do most manual tasks, especially with very different technology like servos, actuators and pneumatics. If we ever get there, the resulting product will be very expensive and still less capable than purpose-built robots. Why buy a $1M humanoid robot that can split logs with your existing $20 axe when fully automated splitters cost tens of thousands?
because the modern world is built around people shaped like people. all the tools and workspaces and interfaces already optimized for it. and that keeps it safe for prople to co-exist. if we start building the world around some otherness, then humans are locked out and obsolete.
It’s 1000x easier to redesign the factory around robots than building and operating humanoid robots.
right, but not every place we’re going to use them is a factory. And it’s possible we’re going to use them in places we’d still like to be human-first because of their critical role, such as i.e. dams
That’s just not true, you’d have to completely rebuild your entire production facilities which would cost more in the long run than taking say… Boston Dynamics Atlas and hooking it up to an LLM trained on a specific task set.
Newer facilities could be built for the future where humans aren’t involved at all, but in the interim making robots that move and manipulate objects like we do is still the better solution.
That’s already the case. Just look at a car factory assembly line, they’re full of robots already but none of them look like humans.
Easier, maybe, but definitely not cheaper.
It is very difficult to make a robotic hand that can operate a screwdriver. If the robot only ever needs to perform one task on an assembly line, just build it with the tools as part of it. Of course, some modularity helps to retool the plant for another product but there are very few cases where a robot needs the versatility of the human hand (maybe bomb defusal?) or body.
yes we have millions of tethered robotic arms in use today with tools as hands. but once an untethered humanoid robot is available on the market with robotic hands, the use cases are infinite.
I don’t think it’s productive to try replicating the human hand accurately enough to do most manual tasks, especially with very different technology like servos, actuators and pneumatics. If we ever get there, the resulting product will be very expensive and still less capable than purpose-built robots. Why buy a $1M humanoid robot that can split logs with your existing $20 axe when fully automated splitters cost tens of thousands?
the estimates are $40k per robot plus ai cloud subcription