Learning vs Following Procedures
Augmenting Safety: AR, VR & AI
These Robots "Learned" How To Move as a Group -- Each One Cannot Move Alone
A procedural program could be written to instruct these robots how to move as a group. But there is one big issue: By design they are not allowed to communicate with each other. Each of these robots only has one sensor and one actuator. The actuator is not enough for the robot to move by itself. And even if a procedural program could be written and if they followed the procedural if-then program flawlessly they are doomed. Why? Because when the environment these robots are in changes (say from a circle to something else) then this group of robots will fail and never succeed in moving again until the environment changes back to the way it was or they are reprogrammed to move in the "evolved" environment.But what if they could use their two sensors to learn about their environment, adapt to each other and their environment and learn to move. That is exactly what this AI experiment is. With just five lines of code these robots are able to learn and adapt. They learned how to move as group completely on their own, without any prior knowledge nor any procedure to follow.
For an Organization to be truly safe from continuously evolving hazards, the Organization cannot rely solely on procedures and policing. It won't work. The Organization must use its given ability to learn. This is why "Learning Organizations" like flight deck operations perform so highly, exceeding the statistical limits of failure in spite of the regular occurrence of human error.
Ref: "Continuous Learning of Emergent Behavior in Robotic Matter" by Giorgio Oliveri, Lucas C. van Laake, Cesare Carissimo, Clara Miette, and Johannes T. B. Overvelde, Designer Matter Department, AMOLF, 1098 XG Amsterdam, The Netherlands, May 25, 2021