3D Scan-Based Slotting Optimization

Problem class

Traditional slotting uses nominal rack dimensions and average item sizes, ignoring the reality that many locations are partially filled, items are unevenly stacked, and actual available space differs from theoretical capacity. 3D scan-based slotting solves the "invisible wasted space" problem by measuring true volumetric occupancy and optimizing at a granularity conventional systems cannot reach.

Mechanism

3D sensors mounted on drones, AMRs, or fixed infrastructure capture point-cloud data of racking faces, measuring the actual physical footprint and height of stored inventory in each location. This spatial data is compared against the location's theoretical envelope to compute true available cube. The resulting volumetric occupancy map feeds an enhanced slotting optimizer that can assign incoming product to locations where it physically fits (not just where the system thinks it fits), consolidate partially filled locations to free whole slots, and identify locations where racking configuration changes (shelf height, beam position) would recover significant space.

Required inputs

• 3D scan data (point clouds or depth images) from drones, robots, or fixed sensors
• Location dimension definitions from the warehouse layout model
• Inventory ledger data (what is in each location, theoretically)
• Inbound forecast (dimensions of expected receipts)
• Slotting rules and velocity tiers (from Slotting Optimization)

Produced outputs

• Volumetric occupancy map (actual fill % per location)
• Enhanced slotting assignments (fit-verified, not just rule-based)
• Consolidation work orders (merge partial fills)
• Rack reconfiguration recommendations (beam height adjustments)
• Space recovery reports (recovered cube by zone)

Industries where this is standard

• E-commerce high-density each-pick fulfillment
• 3PL shared-space warehousing (maximizing billable cube per client)
• Cold-chain storage (space is expensive per cubic meter at -25°C)
• Aerospace MRO parts warehousing (irregular-shaped parts in deep racking)
• Automotive aftermarket (diverse item sizes in limited facility footprint)

Counterexamples

• Uniform, full-pallet-only storage where every location holds one standard pallet to a consistent height — the volumetric variance is near zero and 3D scanning provides no actionable signal beyond what the WMS already knows.
• Rapidly churning flow-through operations (cross-dock) where inventory dwells for <4 hours — by the time a 3D scan is processed, the goods have already shipped. Scanning adds latency without usable insight.

Representative implementations

• Dexory's autonomous scanning robots creating daily 3D maps of UK warehouse operations for Morrisons and other retailers, measuring actual vs. planned space utilization.
• Gather AI's 3D case-counting capability (Ti × Hi estimation from drone imagery) deployed in 3PL pallet warehouses.
• Amazon's internal computer vision systems using depth cameras on mobile robots to map bin fill levels in fulfillment centers.
• Vimaan's PickTRACK system using MHE-mounted cameras for continuous volumetric monitoring during putaway.

Common tooling categories

3D scanning platform (LiDAR drone, depth-camera AMR, or fixed sensor array) + point-cloud processing engine + volumetric analytics module + enhanced slotting optimizer with spatial-fit validation + integration with digital twin model.