> The only non-cgi part of the video was when they showed a regular cart unable to handle plain stairs.
I disagree. The video showing the drone being pushed around by the experimenters is not CGI. It is real footage.
It is weirdly evenly lit, and the cart moves weird of course, so maybe that is why you think it is CGI?
> A place that needs people to carry stuff between levels has elivators and ramps.
Sure. This is more of a “look how clever our controller is” kind of research than a product anyone expects to produce. In other words the outcome here is the learnings from the controller development, not a hovering noisy, windy, dangerous, low-endurance wheelbarrow product.
It is like all those videos where they kick a robot dog and the robot dog keeps walking. They don’t do it because they expect that the future will have a great need for kickable robot dogs, but because it demonstrates how roboust their controller is.
What would be interesting here though is showing the same drone with an off-the-shelf software being pushed the same way. So a software which just tries to stay level and hover at a fixed distance from the ground, but doesn’t have their their clever controller running on it.
Yea I saw the drones pushing eachother. Perhaps I was exaggerating when I said it was all cgi. I just found it amusing that they bothered filming a cart falling down stairs as if its a common everyday problem people encounter and needs some explaining to understand.
To me this is similar to the flying-cars companies.
They all have some weird prototype that can fly for 5 minutes but you never see an actual use or product for them.
I disagree. The video showing the drone being pushed around by the experimenters is not CGI. It is real footage.
It is weirdly evenly lit, and the cart moves weird of course, so maybe that is why you think it is CGI?
> A place that needs people to carry stuff between levels has elivators and ramps.
Sure. This is more of a “look how clever our controller is” kind of research than a product anyone expects to produce. In other words the outcome here is the learnings from the controller development, not a hovering noisy, windy, dangerous, low-endurance wheelbarrow product.
It is like all those videos where they kick a robot dog and the robot dog keeps walking. They don’t do it because they expect that the future will have a great need for kickable robot dogs, but because it demonstrates how roboust their controller is.
What would be interesting here though is showing the same drone with an off-the-shelf software being pushed the same way. So a software which just tries to stay level and hover at a fixed distance from the ground, but doesn’t have their their clever controller running on it.