Automatic forklift - picking a load

Not from the industry, but was involved some 15 years ago in building high bay racks. I think it has to do with the deflections the rack structure is to experience. Every load is different so the deflections are. Pallets close to the upright get less deflections than pallets in the middle. And this deflection is always different, also depending on the load of the neighboring racks when the have a common profile (where the pallets are standing). There is also "double deep" racks where the pallet fork has to reach to the 2nd row, where there is also to consider the deflection of the fork itself.

Hey, I'm new here and not from the industry, but this thread caught my eye. I found it interesting to learn about the problem with picking a load from a rack for automated forklifts. It seems like there are many factors to consider, such as the deflections of the rack structure and the weight and placement of neighboring loads. I wanted to share a helpful link I stumbled upon recently for those who might be interested in forklift training courses. It might be useful for those who are looking to learn more about forklift operations and safety. Just look here for more info. Hope it helps!

pick and place are challenging due to high center of gravity. if things go wrong, forklift and load can tip over.

but picking load is just one part of the forklift activities. the other is navigating shop floor, avoiding obstacles (including humans etc), locating free space and placing load.

there is a lot of things involved in each of them. some of those tasks are solved by AGVs and there are quite few of them out there. so far i only saw one autonomous forklift. i do have pictures and videos of it in action but was asked to keep this under wrap, so don't ask for details. all i can say is that it does all of mentioned tasks and did so without issue for years.

Robots of any type are good a doing the same thing the same way every time. Variability is something they have a hard time dealing with.

With forklift work, often the operator must deal with damaged pallets, non-standard pallets, broken/sagging pallets, loads that are tilted or slumping over the edges of a pallet, etc, and must use creativity and experience to work around the problem. Robots are not good at that.

That's the biggest reason that Warehouse Automation almost always uses high-end, more precise pallets (CHEPs, plastic pallets, etc) that are more expensive, but also more robust, and built to tighter tolerances. WA also tends to have strict rules about how products are stacked and secured on pallets, in an attempt to reduce the variances the automation has to deal with. Similarly, racks for holding pallets are held to firm rules.

Things still go wrong. Loads can still "slump", a rack spot that should be empty might actually have a pallet, or random debris -- things that would be easy for a human to spot, and deal with, are often surprisingly hard for robotics to detect and/or deal with. A robot could easily fail to notice boxes falling off the pallet it's carrying, or a stack starting to sway, which is one reason the vast majority of AGVs/AMRs move quite slowly.

The greater the variety of products, the greater the potential headaches. Single-product warehouses are easier to automate than ones that flow 18000 unique products, all with different box sizes, pallet patterns, wrapping types, and so on. Don't ask me how I know this.

Steering a forklift is something of an art form, as well. Which is why most warehouse robots don't look like forkifts -- building a low robot that centers itself under a pallet and lifts makes the navigation and steering issue much simpler, as does using omniwheels or zero-radius turns. Compare that to the navigational issue of a typical forklift with the steering at the back, and pallets sticking out the front.