Artificial Intelligence in Logistics Engineering

The artificial intelligence (AI) revolution is already here; and the Logistics Engineering Industry is an excellent candidate for a sea-change. This should come as no surprise. Logistics Engineering’s foundational elements make it well-situated to utilize and be impacted by AI; it’s: software-centric, data-heavy, and oftentimes cumbersome. AI offers the opportunity to use software to let the data work for the user, instead of data overwhelming the user. It can help us (the user) make better, quicker, and more informed support system decisions. With the breadth of Logistics Engineering applications, and ever- increasing demands on logistics engineers and analysts, the utilization of AI by the logistics team can optimize their time and facilitate the development of more innovative and effective support systems (that also utilize AI).

The ideas in this article are initial arguments in moving towards the use of AI in responsible capacities within the Defense Industry, and framed specifically within the context of Logistics Engineering activities. Our industry is well-positioned to embrace the wave of innovative ideas flooding from Silicon Valley and the commercial sector. To understand where AI could take us, we must first understand where it came from and where it is now.

A stepping stone to understanding how AI can be used in Logistics Engineering is to understand how a close relative to AI, automation, is currently used. Automation and AI both share a history of being buzz terms. Automation is the precursor to AI. In Logistics Engineering, the use of a common source logistics product data (LPD) database for developing training materials, technical manuals, sparing, and support systems is an example of automation. It requires extensive input data, and defined parameters and processes; but when developed by an experienced logistician, it seamlessly integrates data across a program, and allows the end user to access information they need easily. The end user then reviews that information to make decisions.

In automation, final data interpretation and basic decision-making responsibility lies with the human user (as decisions and/or through the logic for automation software). Through the advent of quantum computing and machine learning, we’ve now arrived at the ability to use AI – where the machine can adjust to changing inputs and make basic decisions. The machine (and/or software) has gained the ability to reason – previously a human-only trait. This frees up the human to focus on making complex decisions and considering factors that do not lend themselves to machine input, while of course monitoring the machine’s decisions.

Commercial products among the Internet of Things (IOT) have already given us insights into the future everyday uses of AI. IOT products exist that respond to our moods, adjust to our daily habits, care for our homes, etc. Products nowadays can observe and react to their environment with very little human input. Driverless cars are no longer a far-fetched notion, but a certain reality. Warehouses where humans formerly ran or drove through aisles, are now robot-filled.

The Defense Logistics Agency (DLA) is finally taking a step towards this direction by teaming with academia to research the use of robotics as part of the DLA Distribution’s Modernization Program. The DLA should absolutely incorporate AI into its operations; but why stop (or begin) at warehousing? AI can be used to benefit nearly every aspect of DLA and DOD operations and program performance and support. At the same time a product shows (or it’s anticipated the product will show) signs of wear-out or reaching its life, a robot could be boxing up repair or replacement parts to be on their way to optimize product availability. Imagine if a system could monitor a soldier’s fatigue (via a variety of indicators, including those requiring perception), respond to the soldier’s personal health data (as opposed to data for the average soldier), and make recommendations to them and their supervisor on changes to make to their activity to remain effective.

Two of the biggest potential benefits haven’t even been mentioned yet: cost reduction and improved productivity. AI can substantially reduce human labor time spent on mundane tasks and basic decision-making. AI-powered support equipment could diagnose, repair, document, and test products, while freeing up the maintainer to address more complex repair tasks. Utilizing AI-powered scheduling systems, more can be done with the same amount of systems via shared use of resources. A present-day challenge is that the maintainer or logistician is inundated with mounds of data. AI can help determine which of that data matters and inform us on trends or actions to take. AI is a true game changer.

How can the Logistics Engineering industry realize the potential of AI? First, address the major concerns associated with AI. A leading concern about AI (and technology in general) is cybersecurity risks and the impact to soldier safety of a hacked system. Any system that is acquired must be secure, and be subject to rigorous testing and monitoring for breaches or access points. Systems must also be easily and safely shut down or returned to human-operator mode if a breach is suspected.

Another concern is the bias that humans trust other humans more than a machine. Would a pilot rather fly a plane that a human or a machine maintainer repaired? It depends, but there is evidence that we hold machines to higher standards than humans, and that machines have to earn our trust more than humans do. For example, driverless car accidents regularly make the news, whereas fender-benders between two human drivers do not. As technology improves, becomes more advanced, will our bias change?

AI-related concerns aside, there are ways that the Logistics Engineering industry can start introducing AI into programs. In fact, we have an obligation in accordance with TA-STD-0017, Activity 7, to consider new technology in performing Product Support Analysis. One way to introduce AI is to push the status quo by questioning “how things have always been done”. Defense contracting program requirements are often derived from previous program requirements, which leaves the window only narrowly open for programs to utilize current or future technologies. Ambiguously-written (or outdated) logistics requirements can be used to provide convincing, proven alternatives (the commercial sector will be a vital resource for alternative development). Programs for initially implementing AI should also be those on which lives do not directly rely, in order to give program offices time to gain confidence in AI’s potential.

Beyond what’s been discussed above, the Logistics Engineering community, and the Defense Industry in general, has another important job associated with AI: demonstrating to US civilians and the rest of the world how to responsibly implement AI. With any new technology, there’s opportunity to use or abuse it. How AI is utilized within US military programs will set precedent. AI is a great opportunity to provide more assistance to our troops when they most need it. Let’s work to set a precedent of using AI to develop better-performing products and support systems that are safer, greener, more maintainable, and cost-effective.

Article Authored by Elizabeth Schwartz