|Sponsor||Plus One Robotics|
|Date||Tuesday, March 21, 2023|
|Time||10:15 AM - 11:00 AM|
As environmentally-minded consumers push for more eco-consciousness in their ecommerce providers, distributors have responded with increased SIOC (Ship In Original Containers) to both reduce waste and lower costs. SIOC, improved custom package creation, tighter advertising connections, and faster cycles have led to extreme growth in the variety of visual appearances of parcels handled in fulfillment and distribution centers.
While the need for increased throughput often drives automation adoption, it’s important to remember that, unlike manufacturing, ecommerce fulfillment and distribution centers are increasingly predicated on this described variability - not repeatability. And the deployed automation solution should reflect this focus by being both flexible and rapidly adaptive.
In this session, Dr. Grollman will discuss the long-tailed and shifting distribution of parcels in fulfillment and distribution centers and how utilizing an adaptable, symbiotic human-ai vision system is the best way to approach the realities of volume and variability. Based on the assumption of change, rather than the anticipation of control, humans and AI working together can handle infinite variety as it occurs.
-Robots need fast, flexible vision perception to 'see' an ever-changing range of packages to perform tasks.?
-It's inappropriate to assume that the input distribution (what sorts of objects the system sees) is fixed, or even fully knowable.
-The goal for warehouse automation should not be zero perceptual errors, but rather a steady-state with some low % of novelty that is handled by remote humans and learned over time.
|Daniel Grollman||Plus One Robotics||Principal Engineer|