On August 28, the Department of Defense (DOD) announced a plan to distribute thousands of autonomous systems across multiple domains over the next two years. The particulars of this so-called "Replicator" project are still vague, however there is one notable detail: the vehicles will be attritable, meaning they can be reused and will be manufactured cheaply enough that they can be left behind if conditions require it.
The opportunities, and challenges, are massive. The intended volumes and variants of Replicator aircraft will require production capacity and flexibility not typically found in the defense industrial base.
For clues on how to navigate this new terrain, it is helpful to look to the automotive industry, which has set the standard for production at speed and scale — and shown how manufacturers can produce innovative variants across a common production platform.
What we know about Replicator
The Replicator project is a long time coming. DOD has seemed to be on the cusp of buying attritable unmanned systems at scale for several years. Replicator in particular is focused on three objectives:
Deploying smart systems in multiple domains. This will rely on a diverse set of variants and intensify the need for tighter collaboration among commercial-off-the-shelf (COTS) unmanned platform manufacturers, artificial intelligence software specialists, payload providers (such as communications), and expert defense platform integrators.
Mass and attritability. This means industry must produce these smart systems not only quickly but at a low enough price point that the DOD will tolerate battlefield losses.
Setting the precedent for how to scale production of defense systems in the future, on top of accelerating production of unmanned systems in mass. This means DOD will be looking to Replicator for lessons learned when planning future investments in large volumes of systems.
Lessons from the auto industry
No industry has more experience producing mass volumes of complicated and differentiated machinery than autos. There are several lessons Replicator participants can draw.
Innovation across a common product platform is the central tenet of any production plan to field a large number of variants. The automotive industry’s rapid, scaled production model starts with its innovation and design philosophy. While commercial vehicles themselves are not attritable, customer preferences change year to year. To keep pace, the industry relies on product platforms that coordinate design and manufacturing strategies to adapt rapidly and at low cost to what customers want, consolidating the development process down to a year or two. These scalable platforms rely on modular product architectures with layered software and include mutual standards such as mounting points for parts common across vehicle families.
The DOD of late is focusing more on modular architectures, spiral upgrades, and collaboration with industry to set common standards. In addition, DOD appears to be taking a family-of-systems approach with phased procurement — investing in mature and readily available systems in the near term while setting standards and planning ramp-up for longer-lead variants with extended range, new payloads, and other capabilities.
Capital-intensive manufacturing investments will require a signal from DOD that robust demand will endure. The auto industry is a leader in flexible manufacturing, automation, and rapid tooling techniques. But many of those investments — such as robotic welding and automatic guided vehicles on the factory floor — are highly capital intensive and are based on volumes that can exceed 100,000 vehicles a year, per plant. It is important to automate where it makes sense while remembering that robots are not the only tool in the toolbox. Digitization of factory floor data collection and analytics, artificial intelligence tools that augment humans for inspection and quality control, and additive manufacturing for new tooling pieces are ways to create sizeable impacts and increased manufacturing flexibility without automotive-like volumes.
In addition, DOD must be aware of industry’s capital allocation requirements and constraints. As seen in DOD’s efforts to accelerate fielding of weapons to support Ukraine and shore up inventories, the defense industry is concerned about investing the capital to increase production and then facing excess capacity in a year or two. The industry will look to DOD for a sustainable, enduring demand signal — without a cliff — to justify those investments.
Original equipment manufacturers (OEMs) should revitalize supply chain strategies to increase their buying leverage, and DOD should focus on materials and components that will be common among OEMs’ unmanned platforms. Benefitting from massive volumes, automakers have shifted from investing in key suppliers for outsourced parts to competitive interval sourcing over the past few decades. Some OEMs have set up central buying organizations that trade and bid for the components used across all the vehicle variants in high volumes, purchases them, and supplies them to each factory. But OEMs may still lack significant buying leverage when demand vastly exceeds supply, as demonstrated by the chip shortage.
Those highly common and potentially at-risk components are good areas for the DOD to make strategic investments — in raw materials, common electronics, and other components that might pose bottlenecks across the OEMs. DOD could also look at a robust government furnished equipment strategy for components like common sensors to create buying leverage and to reinforce the standards the integrators should be following.
Collaborative modeling and simulation environments will reduce test and qualification time and risk. Model-based engineering is critical in automotive testing and validation because OEMs can’t realistically test every combination of mission element and event. Strong model-based approaches for validation are especially important in evaluating atypical mission events in autonomous drive learning, and computer-based decision analysis has become an important part of the software development process.
DOD is also focusing on the modeling and simulation environment, in addition to developing platform-agnostic autonomy software. For example, the US Army is investing in a common modeling and simulation environment for robotic combat vehicles (RCV) to develop and test software across the family of RCV variants.
Tapping “muscle memory” from other commercial industries will help the defense industry develop repeatable processes for scaled production. It is difficult for companies to foresee the issues they will face when scaling to two times or three times current production rates — especially if they have never done it before. Manufacturers that use a strategic launch management plan to systematically roll out product variations can focus only on the change points, improving the time to market.
Automotive OEMs often rely on institutional knowledge to address and troubleshoot unanticipated challenges. Firms can develop the modular platform, processes, and tools — but when things go wrong, teams look to senior engineers and manufacturing experts with decades of insight.