The current-generation Amazon robot assisted warehouses are painfully boring to watch, once you realized how time, energy, and space inefficient they are. Each piece of merchandise has to travel on racks maybe thousands times heavier than itself, at a slow speed, through heavy traffic, while the majority of the warehouse volume (> 80%) is left unused.
Vision: I think most goods can be simply tossed up in the air by robots and be caught by other robots at distances. The future warehouses, which I would like to call them Monkey Houses, should be highly dynamic and densely filled with flying objects. It will improve the throughput of a same size warehouse by more than an order of magnitude and drastically reduce the energy consumption compared to the current systems.
Justification: while humans occasionally use throwing and catching for object handoff (e.g., sports), it is not generally considered as a reliable method, especially when there are multiple objects flying simultaneously. Robots, on the other hand (no pun intended), can be particularly good at this. This includes estimating object motion, performing fast and precise control actions for object catching, simultaneously tracking the trajectory of multiple flying objects, as well as communicating and coordinating with thousands other robots in making plans. With these super-human abilities (i.e., speed, precision, reliability, memory, and communication) of future robots, the engineering trades of future systems design often shift toward counter-(human)-intuitive directions.
More Detailed Vision: each rack in the warehouse will be a stationary robot that can throw and catch objects. Each type of object will have a g-loading rating, dictating how far it can fly in one hop (the packaging of some future goods may have to be redesigned to be better suitable for flying). The goods may go through multiple hops (i.e., catch and throw by robots in between) before reaching final destinations. All object information is shared and an air traffic management system will ensure objects flying pass each other with safe clearances. Like goods, small robots can also be tossed up in the air. They can intersect other flying objects to improve the flexibility of stationary rack robots…
Thinking beyond the warehouse settings, it is conceivable that the main mode of object handoff for robots in the future would be throwing and catching, once the reliability of such systems exceeds human’s capabilities. Compared to the continuous-contact object handoff between two robots, throwing and catching involves much less complex robot-robot interactions and thus is far simpler and robust for robots to perform. This would have many implications to the design of other future systems. For example drone delivery can be performed by throwing packages to balconies equipped with catching robots (or just baskets with nets). Battery changes for drones could be done by simply tossing batteries up and down. Exchange of cargos (and passengers) between two self-driving vehicles on the highway could be accomplished through the air. What other cool applications of robot throwing and catching can you think of?