When implementing real-time location systems (RTLS) in logistics, the amount of data generated can at first be overwhelming. By tracking how assets are moved across a facility, RTLS generates abundant streams of new data, which can exceed a few gigabytes per day, depending on the number of tracked assets and the update rate.
Before RTLS, only limited information was available on where assets were and how they moved in real time at the granular level. Now, you can know precisely how assets are moved, but you may not know what to do with that information.
What’s the answer? Artificial intelligence (AI) in logistics. I’m not referring to AI to automate tasks and processes, but in using AI to generate insights into your process flow by digging into the data. Without AI, logistics operators may be getting little value out of the mountains of data generated by RTLS.
To overcome this, logistics operators must develop a game plan to store, train, process and analyze all this new data. Here’s how.
Building Models
The central challenge of implementing AI in industrial environments is pretty apparent: Once you collect data, how do you use it? In short, you have to sort it to train and build models that produce insights.
Once the data is collected from all sources (e.g., RTLS, cameras, sorting machine, vehicles, etc.) and cleaned, it is divided into training, validation and testing sets. The training dataset is used to teach the model to recognize patterns and relationships within the data.
For example, in logistics, the model might learn to predict delays based on historical asset movement or suggest optimal temporary storage layout based on both real-time item flow patterns and available resources. The validation dataset is then used to fine-tune the model, ensuring it generalizes well and doesn’t overfit to the training data. This step helps balance the model’s accuracy across a variety of scenarios it may encounter in real-world operations.
The final stage involves using the testing dataset to evaluate the model’s performance on unseen data. This provides a measure of how well the model can make accurate predictions or deliver actionable insights in practical applications. If the results are unsatisfactory, the model might require adjustments, such as reconfiguring algorithms, incorporating additional data features or refining the training process.
Challenges To AI Adoption
Even if you follow the aforementioned steps to the letter, several common pitfalls can derail AI implementation.
1. Unreliable Or Untrustworthy Data Feeds: Poor-quality data leads to poor-quality results. Or as the old adage goes, garbage in, garbage out. Without clean, consistent and trustworthy data, AI models cannot deliver consistent and accurate insights. AI models are highly dependent on the data they are trained on and fed, and if the data is inaccurate, inconsistent or incomplete, the resulting insights will be flawed. For example, malfunctioning sensors, unreliable asset location or gaps in data collection can produce incorrect conclusions about asset movement, leading to questionable insights.
2. Lack Of Expertise Among Employees: AI and ML are powerful tools, but without the right expertise, their potential often goes untapped. Many logistics facilities lack team members with the skills to implement, manage and interpret AI models effectively. This can result in a range of challenges, from inefficient use of resources to an inability to troubleshoot AI-related issues.
3. Blind Faith In AI-Generated Analysis: While AI can offer valuable insights, relying on it without understanding how it arrives at its conclusions can be a major pitfall. Blind faith in AI can lead to misguided decisions, particularly if the underlying data or models are flawed. For instance, if an AI system suggests optimizing inventory placement based on incomplete data, the result could be increased inefficiencies rather than improvements.
Solutions
To successfully integrate AI into industrial operations, it’s essential to address these challenges head-on, starting with data quality. Reliable, consistent, high-quality data provides a solid foundation for process analysis and optimization.
Often, it is very valuable to also capture the level of noise or confidence (e.g., through variance) of the measurement data itself. Many organizations use a data lake to store vast amounts of raw data in its original form, allowing them to sort and analyze it later. This approach ensures that valuable information is preserved and can be leveraged for deeper analysis when needed.
Equally important is the need for expertise in AI and ML within the workforce. Hiring specialists with experience in these fields can help an organization implement AI effectively, but companies should also focus on upskilling existing employees.
Training programs, workshops and collaboration between logistics and AI teams can foster a deeper understanding of AI systems. This not only improves implementation, but it also ensures that the workforce can actively contribute to AI-driven initiatives, making the technology a practical part of daily operations.
Finally, it is vital to address the issue of blind faith in AI-generated analysis. Leveraging explainable AI (XAI) can be a crucial step to ensure that operators understand how AI arrives at its conclusions. XAI provides transparency into the decision-making process, explaining the logic behind each recommendation and highlighting the data used. This allows operators to evaluate the credibility of AI-generated insights and make informed decisions.
By providing transparency, XAI fosters trust and encourages broader adoption of AI within the organization. It also helps overcome employees’ reluctance to rely on AI, as they can see the rationale behind its suggestions and feel more confident in its recommendations.
Final Thoughts
By focusing on data quality, expertise and transparency, logistics operators can effectively leverage AI to turn a vast amount of raw data, including RTLS and location information, into actionable insights.
When implemented strategically, AI can create more efficient, agile and resilient supply chains, positioning logistics operators for long-term success in an increasingly competitive market.
This article was originally published on Forbes.com.
