Originally published at: https://boingboing.net/2019/07/16/robots-made-from-tree-branches.html
…
Success?
(I, for one, welcome our new Robot/Stick Overlords.)
I know I’m thinking anthropomorphically here, but I feel even sadder for physically challenged tree robots forced to learn to walk with difficulty than Boston Dynamics’ abused robots.
Robots made from tree branches learn how to walk
It’s like robot QWOP
Awww. It’s like watching baby Groot take his first steps.
I find the learning process goes much better when I use carrots instead of sticks.
It would be interesting to know how thoroughly they investigated the losses due to following a ‘local minimum’ with their DDPG/“Deep Deterministic Policy Gradient”. That is, there could be a much more productive gait which isn’t discovered due to settling on an initial slightly productive mode plus incrementalism from that point only. And/or is there any improvement towards discovery of a (more) ‘global minimum’ beyond genetic algorithms and other ‘ant colony’ multi-agent approach, or simulated annealing?
The larval stage of a Survival Research Labs creation.
March of the Ents.
You said " ImatellaNobody"…
Another staggering achievement of Twig TechnologyTM.
I hope they stick with this.
Unless it leads to Robot Triffids.
That would be bad.
Triffids… now that would be a cromulent use-case for the flamethrower drones.
Sometimes BoingBoing is quite prescient…
This topic was automatically closed after 5 days. New replies are no longer allowed.