A modern warehouse no longer ships single-product pallets. Orders mix dozens of items, their profile changes hour by hour, and at the end of the line someone has to put all those boxes on a pallet — stable, dense and truck-ready. Done manually, it is hard, inconsistent work. Done with a rigid system programmed for fixed patterns, every new product means reconfiguration and engineering time.
Mixed-case robotic palletizing solves the problem differently. Instead of fixed rules, it uses real-time perception and adaptive decisions. ZEDlog integrates this capability through AnyStack, the system built by our technology partner Progressive Robotics.
What "mixed-case, on-the-fly" means
Two words carry the difference. Mixed-case means different boxes on the same pallet — varied sizes, weights and packaging. On-the-fly means the robot decides where each box goes at the moment it arrives, without knowing the product order in advance.
That matters because in practice products never arrive neatly sorted. They come in mixed, unpredictable streams that shift through the day. A system that needs a preset order has to create it, with extra equipment. An on-the-fly one does not.
The loop that does the work: perceive, decide, move
The differentiator is not a single feature. It is the closed loop running in real time:
- Perceive. A 3D camera scans and measures every box on arrival, with or without a barcode, and detects damaged or shifted boxes.
- Decide. In milliseconds, the system evaluates where to place the box, applies stability rules — heavy low, overhang limits — and verifies stack integrity before moving the arm.
- Move. The planner generates a collision-free trajectory optimised for the palletizing cycle.
The software works out the best way to build each pallet on its own: it chooses the layer pattern, adapts to box sizes and gripper limits, runs a stability analysis and estimates how much each box can bear without being crushed. You do not have to change your process to fit the robot.
The brain: the motion planner
The part that really matters sits in software, in the motion planner. It calculates how the arm moves from one point to another without hitting anything — and, more to the point, how fast and on what hardware. Here is what Progressive Robotics says about it:
- Speed. Collision-free trajectories computed in milliseconds, which lets it generate the path in real time. Without that, mixed-case palletizing would not be possible.
- Runs on a CPU, on-premise. All the calculation runs on a standard industrial PC, with no expensive graphics cards. Lower cost, predictable timing, simpler integration.
- Robot-agnostic. Validated on 4-, 5- and 6-axis arms from various makers, plus linear axes and custom grippers. No lock-in to one supplier.
- Singularity-free and collision-free. It avoids the configurations where a robot gets stuck, and it accounts for dynamic obstacles too — including boxes already placed on the pallet.
In short: fewer stops, less manual tuning, more uptime.
Why we drop the sequencing
The industry ran for a long time on sequenced systems: buffers and sorters that arrange products in a specific order before palletizing. They work when the flow is predictable, but they bring six classic problems — buffering, recirculation, floor space, control complexity, maintenance and high upfront cost. Plus one more place the line can stop.
The on-the-fly approach takes all that infrastructure out of the equation. The robot plans motion and placement together, straight from the random flow, and delivers a stable, truck-ready pallet. Less equipment to manage, a simpler flow.
What it looks like in your warehouse
AnyStack comes as a complete cell with a single point of contact: robot, software, safety system, commissioning. It handles both inbound palletizing (building single-product pallets from unloaded containers) and mixed outbound palletizing, onto pallets or roll cages. The interface is built for warehouse operators, with no robotics knowledge required.
ZEDlog's role is integration: we design how the cell connects to your physical flow and to the WMS, size the throughput and commission the system. It is the same logic we apply to conveyors or sorting — we start from the operation's data, not from a catalogue. See also the guide to warehouse automation, which palletizing is part of.
The profile that fits best: eCommerce, retail and 3PL operations with many SKUs and orders that change often. Exactly where flexibility matters more than a fixed pattern.
If you want to see how it would look on your real products — sizes, weights, pallet type — we start from a conversation about your flow. Write to us and we take it from there.
