Engineering & simulationAmby Stack
The system uses a reinforcement learning model trained in simulation to autonomously determine optimal stacking patterns for various box types and weights, replacing manual programming of pallet locations.
“the unique thing that that Amby has done is they have trained the AI in simulation with a reinforcement learning model that's a, Jeff describes as a game. right? So the robots rewarded when it does something well, right? in this gameified simulation, until it figures out the right way to actually step all of the different types of boxes that might come down.”
Speaker 2 · Hear it at 14:37