In recent year’s world has seen multifold increase in the e-commerce and online selling, to cater to these demands companies have setup multiple warehouse’s across the country and across the world. With the increased warehouses, stocktaking in these warehouses has become one of the most important processes. Accurate stocktaking or stocking is an art and a science, it requires experience and precision. When done right, stocktaking leads to efficient picking, the ability to meet demand, identify items reaching expiry or are obsolete, adjust pricing, identify stocks for flash sales, order replenishments and detect discrepancies that could be the result of theft and pilferage. Stocktaking is therefore a major event in large warehouses. For warehouses that run into millions of square feet it becomes necessary to call in additional manpower and keep the warehouse shut for four or five days while stocktaking is carried out each month or quarter. During this period, there is no inward or outward movement of goods, leading to direct and indirect losses. In fast-moving business environments, long-drawn stocktaking is untenable. This led us at Wipro to develop a faster and accurate solution which ensured lower cost of ownership, and a more reliable way to manage stock. We built a drone-based solution that replaces manual stocktaking – and it can even do stocktaking when everyone has left the warehouse for the day.

The science and technology of stocktaking has seen some improvement over the years. Modern warehouses have teams climbing up three, sometimes four, levels in a stocking warehouse and use a mobile app/ readers to record items on each pallet. While this helps, it is still laborious and error prone.

Using advanced stock technology to create a solution

We created an affordable drone-based solution for a FMCG client who wanted an automated solution for stocktaking. One option we had was to use RFID stickers on pallets/ containers/ boxes. RFID stickers can be easily scanned and each item recorded. However, RFID stickers are expensive. They cost around INR 7 (~10 cents) each, plus there is the additional cost of RFID printing and scanning equipment. That works out to more than INR 700,000 (~$10,000) for every 100,000 items. FMCG is a thin margin industry. It cannot afford the luxury of RFID. This is how the drone solution emerged.

In its simplest form, the solution works like this: We equip a drone with a camera, and the drone will navigate inside the warehouse autonomously, it can fly within the aisles and capture images of the pallets. A set of deep-learning based box counting algorithms examines the images and produces a record of inventory.

At the click of a button, a warehouse manager can send an autonomous drone and complete the task without paper or additional manpower. Each pallet has a sticker. By reading the sticker, the system knows what the pallet contains. Accuracy greater than 98% has been recorded.

The drone and its control system can be deployed just by simple commands. No one needs to climb boom ladders (by using drones we prevented a few accidents as well!). The drones use visual navigation to find their way since GPS signals are not available inside a warehouse. Once done (or if they are about to run out of charge), the drones return to the charging station and resume the task with a battery replacement or a re-charge.

Bringing more decision power into the hands of managers

The images collected by the drones are pushed to a vision server where the images are processed to count the number of boxes and a report is generated. Using the data, the warehouse can easily identify the number of cases for any product SKU and their exact location in the warehouse. The drone can see boxes in the front side and boxes behind them will not be visible. But algorithms are able to estimate the total number of boxes in each pallet accurately, if the proper pallet storage process is followed.

The inventory report generated by the system can also be integrated into any incumbent warehouse management system solutions. Here, historical data is available to make a variety of analytics-based decisions. The application can be accessed over a mobile phone, making it easy for warehouse personnel to manage inventory.

Some benefits of the stocktaking drone are evident:

• There is no need to shut down the warehouse during stocktaking, leading to cost savings
• The drone can be deployed day or night, leading to time flexibility and higher efficiency
• Manpower is not required (except the operator), leading to zero productivity loss
• Stocktaking can be done in more frequent cycles than weekly or monthly
• Stocktaking can be scheduled or ad hoc, leading to flexibility
• Warehouse stocks can be matched instantly with ERP data for better order fills and happier customers
• Warehouse managers can be equipped for highly accurate decision making (related to discounting etc.)
• Inventory can be checked selectively by picking an area on the warehouse map or by value of product, expiry date, etc.
• A simple solution which requires minimal training

A system so good, you’ll want to take it for a spin

At the moment, the system delivers 98% accuracy for counting. For the balance, 2% the system depends on user inputs. At the click of a button, the user can approve the count or provide the right value.

As of now the drone can fly for 21 to 25 minutes on a single charge which typically allows them to cover many racks or more than one aisle, depending on dimensions. We are having plans to increase the dron flight time in the coming future.

Warehouses can use the drones in two business models. They can invest in drones upfront and run the operation independently or they can use our as-a-Service model and use the drones once a month or quarterly as required.

For warehouse managers struggling to solve stocktaking, auditing and reporting processes—without disrupting operations—drones can be a cost-effective game changer.

Anandaraj is a DMTS - Senior Member at Wipro Limited where he is leading the Drone based solutions and Hyper Spectral Imaging research initiatives. He has over 21 years of experience in Solution development and Research in various industry domains including Healthcare, Energy & Utilities, etc. 

He has published research papers, spoken at international conferences and filed/granted patents. He had also participated in the working groups for defining the standards and solutions for healthcare (M2M for Health), IoT (IEEE P2413) & Smart Agriculture. He is actively involved in academia connects and invited talks. He is presently serving as a member of industry advisory board and Board of Studies for academia.