Introduction

Machine vision—traditionally thought of as a technology to spot defects or out-of-spec parts–has found a new use: pinpointing the location of materials in warehouses.

Open storage areas are present in almost every factory and warehouse, yet few methods exist to precisely determine and record material locations. Consequently, materials are misplaced, inventory accuracy suffers, and labor is wasted locating lost goods. By applying contemporary machine vision technology, Sky-Trax Inc., which develops technology to locate materials in warehouses, has developed a practical new solution to this old problem.

The Problem

Many current locator methods fall short of providing sufficient accuracy to locate a single pallet amidst thousands of others. For example, floor-painted aisles, overhead bar code placards, and sector markers can identify approximate location. Bar code labels or RFID transponders embedded in the concrete floor improve accuracy and automation readiness, but even these suffer from missing or imprecise location data due to sparse placement and the tags' vulnerability to physical damage and obscuration.

The Solution

Placing visual location markers overhead in the warehouse ceiling, and utilizing machine vision sensors to view upward and interpret the markers, has been a breakthrough concept. Marker panels are fabricated by Sky-Trax Inc. from sturdy but lightweight netting, and an array of printed position markers are attached at predetermined intervals. The netting with position markers is suspended above the area where items are to be stored or vehicles are to be tracked. Each position marker contains a coded symbol that uniquely identifies its place on the net which, after installation, is mapped to a corresponding floor location.

Embedded software running in commercial machine vision equipment performs the primary task of locating the coded symbol within the field of view, decoding it, and relaying the decoded data to a controller. The vision system, powered by DVT, determines the position marker's exact location and orientation within the field of view, and calculates its corresponding position in "real" space. All of this occurs within a few tens of milliseconds, yielding position update rates of several samples per second. Serving much like a global positioning system (GPS), the local positioning system (LPS) gives accurate position information that can be used to track stationary or moving objects. With this capability, a user can know in real time the exact location of equipment, materials, or individuals to an accuracy of plus-or-minus a few inches.

Sky-Trax incorporated a vision system on lift trucks to help operators locate stocks.

With the controller wirelessly receiving position data from a fleet of lift trucks, the motion and activity of each truck can be monitored. A graphic display on the controller console shows animated truck positions overlaid on a drawing of the warehouse floor plan. The lift truck operator views a GPS-like screen that shows his position in the warehouse, vehicle heading and velocity, and the location of other trucks. Coupled to a warehouse management system, the LPS reliably captures the location of each pickup or put-away with high accuracy and directs target pick and drop locations.

Conclusion

The Sky-Trax system with machine vision is in the beta-test process now. The company is also exploring other applications beyond forklift trucks.