Note how the robots are lined up and waiting most of the time, meaning that the bottleneck is in human workers. I'm guessing it's not long till they are completely eliminated from the order assembly.
Your guess would be wrong. Dynamically perceiving, picking and placing irregular objects is really hard and completely unsolved. Obstacle avoidance via sonar & lasers and reliable bar code reading have been solved for a decade.
They didn't create an advanced robot. They created an advanced system that works really well in whole.
> They didn't create an advanced robot. They created an advanced system that works really well in whole.
Of course, I understand that.
But if you watch the video again, you may notice that a vast majority of items being shipped are .. boxes, not irregular shapes. Determining if the order consists entirely out of box shapes is straight-forward. And since the packaging needs not be optimal in terms of space, just not overly wasteful, so there is a lot of potential for further automation.
True....and cardboard boxes often come with the flaps connected when they are folded...I see no reason why a robot couldn't unfold the appropriate size box for an order, leave the top flaps open (but still connected, giving at least 50% more volume than a sealed box), then a sophisticated warehouse like amazon has could likely easily treadmill individual the items into the box (in proper order according to item size, etc), scanning them as they go in to ensure success, then the fine tuning of packing could be done by a human. Some orders might not qualify for this level of automation, but thats pretty easy to figure out.
I would imagine Amazon might already be doing this, anyone ever seen a video??
I meant regular in the the industrial manufacturing sense. There are, at least, a few different kinds of boxes - and they are made of a warpable material. That means you need to perceive the environment, no just assume you know it. That makes it hard.
> Dynamically perceiving, picking and placing irregular objects is really hard and completely unsolved
wrong. it is completely solved, but simply too expensive. in a couple of years this technology will be cheap and replace most human pick & place processes. The same thing happened during recent years and will continue with exponential groth.
