Xrobots.org mission: democratize robots.
The how is to use hardware, customers, and VUCA environments to create incredible AIs.
Hardware first:
AIs trained by experts suck because of biases, data shift, underspecification - as Rodney Brooks put it - because of pretension. What if we put a million physically capable but computationally incompetent robots into the world and let them bump around stupidly so they become smarter through experience? Standardized, modular hardware, like Clicbot, but with arms, end effectors, and cargo capacity so the robots have the capacity for utility instead of just novelty. Facebook works because of the network effect - robots need the network effect. There aren't enough robotics postdocs in the world to make that happen, so put hardware into common peoples hands instead. How did ASU become #1 in innovation over Stanford, MIT, Harvard, etc? By flipping the ivy league model on it's head - quantity over quality - #diversityandinclusion. 1% or less of the population can code. Half of the U.S. population can't read at the 8th grade level, but nearly everyone can tell a robot what they want, show it how to open a door, or mold its effectors around an apple. Democratizing robots achieves the n size needed to create robust AIs.
Customers:
Rodney mentioned that we must not put cognitive load on customers. For a vacuum or box packing robot I agree, because those tasks are dull and therefore draining. Yet, if the UI was gamified and socially connected, customers could have fun training their robots alone or with friends through voice, demonstrations, and puppeteering, and the cognitive load would be accepted because play is energizing. If customers can see the potential in the hardware because it has arms and effectors but customers know the robot must be trained to do anything, there is no expectation violation, there is no buyers remorse, and if the robots are cloud connected and learn from every human interaction across all robots then it gets smarter with "time" (hidden effort from other people). Customers will train robots unlike researchers - they'll be careless, cruel, foolish, risky, etc. - robots need customers.
VUCA environments:
Labs suck. Volatile, uncertain, complex, and ambiguous environments are where it's at. What if we created challenges (like DARPA's subterranean challenge) with rewards like a video game and people will stress test their robots in the worst environments. Tell people to work in the mud and they'll groan because it's dirty. Tell people to play with their robot in the mud and earn 1000 points and a badge and they'll pay you for it and have fun doing it (e.g. Boy Scouts).
Business model:
Who pays for all this hardware? A startup couldn't survive it. VCs won't let that much risk happen. A non-profit like Arduino and Openrobotics is the only way to feasibly do it, which is why I bought Xrobots.org.
I'm searching for your reactions to these ideas, a technical co-founder, board of directors members, and technical advisory board members. We have 2 advisory board members already. Please help us to further this effort to democratize robots.