SPONSORED CONTENT BY
Distribution centers provide a controlled environment that is ideal for testing and proving complex technologies like drones and robots. That's also one reason why DCs are experimenting heavily with Artificial Intelligence (AI).
An independent research survey commissioned by Lucas Systems found that the majority of companies are already using AI in their warehouses and distribution/fulfillment operations. The survey also revealed that operators view cost, complexity, and lack of understanding of how to use AI as key impediments to further investments.
In reality, AI will make it easier and less costly for DCs of all sizes to address warehouse optimization challenges like slotting and workforce planning. And successful use of AI will not require massive investments in data science departments. Here's why.
Good data is a key to effective AI , and DCs are a good environment for collecting and aggregating historical and real-time data. AI is also a natural fit for DC operational challenges that previously required highly-engineered expert systems that are costly to implement and maintain.
AI and machine learning-based solutions reduce those obstacles, and they give DCs better results than current resource and inventory management approaches that rely on Excel, inherited best practices, or simple rules-based decision-making. AI is making advanced optimization practical for smaller operations, and more flexible and cost-effective for larger facilities.
Lucas has identified five key applications for AI in the warehouse today.
Proper product slotting impacts labor productivity, throughput, and accuracy, but doing it well isn't easy. Slotting is both a combinatorial optimization problem (many input factors to consider) and a multiple objective optimization problem (with many goals, sometimes competing). In addition, there are thousands of products and product locations (slots) to consider, and those products and locations may change frequently. Traditional slotting solutions require customized models and extensive engineering, measurement and data collection, both to install and maintain.
AI eliminates much of the engineering work and manual warehouse mapping and data inputs required for traditional slotting systems. AI-based software can learn the spatial characteristics and travel time predictions required for a slotting model based on activity-level data captured in the DC. And the learned model will adapt as conditions change, providing continuous optimization.
Optimal labor allocation is essential to ensuring orders get out on time while eliminating overstaffing and understaffing. In many DCs, supervisors make staff allocation decisions throughout a shift based on the volume of work, deadlines, and current and expected productivity. Good decisions require good data and accurate predictions, which today are often based on each manager's individual experience and skill.
To improve results, machine learning can be applied to predict labor requirements and work completion times. An AI solution can also run simulations to determine how to best complete the work, avoiding delays and ensuring the most efficient use of labor.
Labor management systems using Engineered Labor Standards (ELS) have been around for years. AI can eliminate much of the labor-intensive data collection process required with ELS-based performance management, using learning algorithms to predict the time required to complete tasks.
AI algorithms learn based on real-world performance data collected from within the operation, taking into account a multitude of variables (user, work type, work area, starting travel location, ending travel location, product to be handled, quantity to be handled, etc.). The predicted results and expectations are more accurate and the ML models adjust when operational changes are introduced.
Warehouse workers spend much of their workday traveling within a facility, making travel reduction a key to improved productivity. Automation and robots each eliminate travel, and AI can be used in areas where automation alone is not enough .
AI and machine learning systems use large amounts of process data to 'learn' how to balance priorities and reduce travel through intelligent order batching and pick sequencing. The systems take into account common congestion areas and slow-moving routes. Many DCs have achieved 2x productivity gains in piece picking applications using AI-based travel reduction, and even case pick to pallet operations have demonstrated 20-30 percent productivity gains.
The same tools used to optimize travel for workers can apply to orchestrating people and autonomous mobile robots (AMRs) in an order-picking process. In most pick-to-robot systems today, the robot system optimizes and directs the robots to a location, and a nearby worker delivers one or more picks to the robot based on instructions on a tablet mounted to the machine.
An AI-based execution system can orchestrate and optimize for both the robots' and the pickers' time – while also providing means to direct workers independent of the AMRs (using wearable mobile devices rather than robot-mounted tablets). Machine learning algorithms predict where the robots and pickers will be located at a given time, and other algorithms provide input to intelligently organize and sequence the work among people and robots.
In the survey mentioned earlier, the cost was seen as the biggest impediment to AI adoption, and 8 in 10 of the respondents also said their organizations need a better understanding of how AI can be used in the DC.
As outlined above, AI has the potential to reduce the cost and manual engineering time and effort required to implement a range of DC optimization solutions, from slotting to labor performance management. What's more, these new AI-based solutions do not require that companies develop extensive in-house AI expertise.