Future Land Warfare Collection 2021: Joint Logistics Through Robotic and Autonomous Systems - Opportunities and Risks

The question is not whether the future of warfare will be filled with autonomous, AI-driven robots, but when and in what form.[1]

Robotic and autonomous systems, colloquially known as RAS, offer significant potential in the joint force logistic effort that extends from the national support base along lines of communication to frontline use. Industry has been employing automation and autonomy for some time, yet Australian Defence Force (ADF) adoption of such technologies has been evolutionary and somewhat halting. The recently announced Force Structure Plan and Defence Strategic Update highlight the increased prominence of RAS, with Army being allocated up to $11 billion for future autonomous vehicles, Air Force investment in teaming air vehicles of up to $4 billion, and Navy committing to an uncrewed suite of capabilities, particularly for sub-surface warfare. As further demonstration of its commitment, Army requested and received funding and workforce to establish a coordination office for the implementation of RAS.[2] This is recognition of the rate of change of technology, as articulated in the Australian Army’s Accelerated Warfare statement:

… we must prepare for an accelerating environment. Future warfare, in certain parts, will be fought at the speed of machines with success belonging to the side who can adapt the fastest.[3]

Army released its Robotic and Autonomous Systems Strategy in late 2018 to outline the value it seeks to gain from the adoption of this range of technologies. The Australian Defence Force Headquarters (ADF HQ) has yet to issue its RAS concept, although it has been crowdsourcing ideas and thoughts under the ADF—Concept for Robotics and Autonomous Systems 2040 (ADF-CRAS 2040) banner. A low-risk area of significant potential for RAS is joint logistics, especially in the deployed environment.

This essay outlines the opportunity that RAS offers in joint logistics and some of the challenges that may ensue, and finally offers some suggestions for a way forward. Overall, the ADF is at risk of not realising many of the benefits of automation and autonomous systems because of conservatism, process rigidity, a lack of cooperation and, particularly, the want of a coordinated joint approach.

Definition of RAS

Before exploring this topic, it is valuable to consider what is meant by RAS. The Army RAS Strategy describes it as ‘the application of software, artificial intelligence and advanced robotics to perform tasks as directed by humans’.[4] Likewise, it has also been described by Paul Scharre as ‘a machine, whether hardware or software, that, once activated, performs some task or function on its own’.[5] The key is that a RAS can be physical or non-physical and that it fulfils a function without requiring human input. Levels of autonomy and functionality are variable, with graduated levels of human input or supervision. This ranges from remote control, which is how the military traditionally uses Uncrewed Air Systems (UAS), to semiautonomous systems where relatively simple inputs enable control, to full autonomy. Full autonomy is currently employed in some manufacturing processes and in the mining sector, where humans merely supervise autonomous dump trucks from thousands of kilometres away.

The absence of an internationally agreed scale of autonomy confuses discussion and complicates analysis. This labelling problem is compounded in the military, as capabilities are often seen as a ‘system of systems’—where labelling a system as autonomous when not all subsystems have autonomy can be especially misleading. This definitional challenge is worthy of a study in its own right, but for the purposes of this essay, RAS will be considered broadly as systems (both hardware and software) with variable degrees of self-determination and, in some cases, robotic manifestation.

Opportunities Provided by RAS

The potential for RAS to benefit logistics has been highlighted in the UK, where it has been identified that ‘[s]ustainment will be improved ... by improved stock and platform monitoring and anticipation; but also by automated logistic delivery’.[6] It is helpful to note that logistic systems are configured to overcome two key problems—time and volume. The requirement to have the correct commodity, in the right quantity, at the right place and in a timely manner drives the logistic structure to support an operation.

In the Army context, the time and volume problem has been resolved through the echelon system. Logistic elements to support the fighting echelon (F echelon) are arranged through increasingly large elements the further one moves away from the F echelon, where it is considered that risk is lower. The echelon system holds essential combat supplies[7] very close to the point of need so that the F echelon can be replenished rapidly, highlighting the criticality of time. Addressing the problem of volume, the echelon system holds increasingly large quantities and large physical items away from the point of need. This means they may be more secure and also reduces the requirement to relocate this larger stockpile as a result of the ebbs and flow of battle. It does, however, make timely delivery more difficult.

This echelon system has been in place for over a century, has served Army well and is akin to the approach of Navy and Air Force, albeit on a different scale. Similarly, high-value but low-population items such as aircraft engines or missiles are held away from threat where they can be safely stored, maintained or assembled. This is due to the complexities involved in relocating them, sensitivities with transportation, and the requirement to carefully manage the commodity. This approach does, however, pose a challenge of time, as the item often is not readily available at the point of need, or a receiving element has to be moved to a specific loading point—for example, to reload missiles or torpedoes. These challenges of time and volume could be mitigated, to a degree, by RAS.

There is an opportunity to address time and volumetric limitations through significantly enhanced logistic situation awareness, monitoring and artificial intelligence (AI) assisted decision-making. This would require more than simply an enhanced recognised logistic picture; rather, it would require a system that fuses logistic information, real-time usage monitoring and an understanding of future intentions. Such a system would be able to not only identify what is needed and where but also recommend, plan, and deliver the commodity in a timely manner. This would reduce stock waste and avoid unnecessary logistic movement.

There are already nascent systems available, and many airlines and commercial haulage companies use an automated approach to pre-position parts to manage engineering demand and fault resolution and to maximise the availability of assets. Indeed, the F-35 comes ‘bundled’ with an autonomous logistic information system, although reportedly the system suffers from lack of performance and user confidence. As a result, the US military is changing to an operational data integrated network (ODIN) system—highlighting that the concept is sound. Yet the potential for such approaches extends beyond engineering and into high-volume and urgent combat supplies. A synthesis of monitoring usage, predicting the point of need and arranging delivery could enable a more agile and targeted system. In the land domain this could allow Army to reimagine the echelon system.

To address the logistic challenge of time requires examination of the rapidly changing range options for distribution. In Army’s case, delivery of combat supplies to the F echelon requires the commodity to reach the force on land through a network of surface (motor transport or watercraft) or air (rotary wing, air delivery or aeroplane) means. The environment, distance, threat and weather all constrain options for where those supplies can be positioned. Given that the planning norm for A1 echelon[8] (immediate stocks) is to provide replenishment within 15 minutes, a ground transport solution has to be within 5 kilometres to meet such a requirement, exposing it to both direct and indirect fire attack. RAS offers some new opportunities here. For example, aviation is often a scarce resource for logistics, given other priorities such as casualty evacuation or troop movement. However, a ‘heavy lift’ UAS could provide a viable tactical resupply alternative. An air delivery solution able to fly at 150 kilometres per hour could still meet the response time while placed almost 40 kilometres away, well out of the range of most artillery. Furthermore, this approach means that the individual A1 echelon stockpiles could be significantly reduced, addressing some volume concerns by reducing what has to be held ‘just in case’.

A heavy-lift UAS therefore allows the Army to reconsider the echelon system in its current guise and provides opportunity for fundamental change. Furthermore, the capability to lift and move meaningful volumes of freight autonomously by air is also a joint opportunity. A heavy-lift UAS could be especially effective in addressing the critical ship-to-shore connector role for a landed amphibious force ashore and so enable sea basing. Such platforms could also be an effective naval ship-to-ship connector managing commodities across a naval task group. Finally, on a deployed air base a heavy-lift UAS might enable better linkage between flight line and storage sites. In this case, commodities like ammunition that impose significant safety constraints, or electronics that require climate control and have a significant electrical footprint, can be stored away from habited work space or aircraft and be rapidly brought forward as needed.

In addressing combined time and volume challenges, autonomous vehicles also offer potential, especially for the land environment. One of the limiting factors in the logistic system is the ability to keep delivery assets moving. Trucks are often crewed by two personnel in order to maximise how long the truck can operate and survive in the environment. Many nations do not crew beyond that; therefore there is a limitation on the time for which a delivery asset can be moving freight.[9] Of course in times of crisis, truck operators could be required to continue well beyond what might be deemed safe in peacetime. However, assuming a two-person crew, it can be concluded that a truck is only operating at 75 per cent capacity as it is immobile for six hours a day.[10] The corollary is that is that Army needs 25 per cent more trucks in the fleet to mitigate this downtime.

A RAS leader-follower capability offers very significant logistic opportunity, since it requires only the lead platform to be crewed, while the followers are uncrewed and autonomous. If the lead vehicle can carry multiple passengers—for example, a Bushmaster—this enables continuous crew rotation. Consequently, apart from refuelling, the platforms can operate constantly for as long as the maintenance allows—conceivably up to 10 days. This approach offers at least a 25 per cent increase in the volumetrics of what can be moved without growing the workforce. Furthermore, future cargo vehicles designed as followers do not require a cab for the crew—further increasing the load carriage capability by between 1,000 and 5,000 kilograms depending on truck type. The Australian Army has begun to experiment with a sovereign leader-follower capability through a research agreement with Deakin University and will test this hypothesis over the next few years, including on civilian roads.[11]

Autonomous ground transport also addresses some of the risk associated with operating in the land domain by automating delivery. Resupply is a critical vulnerability of a deployed force, and reliance on land lines of communication has been a vulnerability since ancient times. It has been reported that a little over half of US military casualties in Iraq occurred from attacks on land transport.[12] The US Army has an ambitious autonomous truck program and intends to have 300 deployed by 2025[13] that will operate in high-threat environments. The UK is also participating in the program[14] but has a different concept of employment. Based on experiences in Helmand province in Afghanistan, the UK is less convinced about the capability of autonomous trucks to deal with the uncertainty associated with complex environments. Instead it sees autonomous trucks being employed on longer duration, less demanding resupply roles, thus freeing human workforce for the tactically complex. This divergence of views highlights one of the challenges of RAS: sensors and processing still struggle to deal with the unusual and complex, especially on the ground in populated areas, and thus there is gap between aspiration and reality. In contrast, for the Air Force, operating in an environment devoid of the local population, such as within a deployed air base, autonomous ground platforms afford efficiency gains, as Project Kelpie has demonstrated, delivering spares without needing the technician to leave their station. An added bonus is that autonomous ground transport reduces the risk posed through fatigue and the incidence of humans falling asleep on long drives—highly apposite for Australia.

Challenges to Adoption of RAS

A number of challenges are associated with the military adopting RAS technologies. There are risks such as networking, vulnerabilities to cyber-attack, and uncertainties about the ability of delivery systems to perform in all weathers, day and night and against a range of threats. A bigger challenge is cultural inertia. Theo Farrell states that ‘military organizations, as socially conservative and closed communities (not unlike religious orders), are especially disinclined to innovate.[15] This is to be expected, as often technology is disruptive; it challenges our worldview and therefore potentially destabilises the status quo internally. It also poses a risk that, if realised, could mean a military fails in the next war. Cultural inertia reflects the historical paradox that in peacetime militaries laud caution, accountability and a deliberate approach, whereas in conflict they require risk-takers, mavericks and disruptive thinkers to drive adaptation and create operational opportunity.

To reconcile these conflicting cultural needs, visionary and empowered leaders are needed to drive innovation from within; it cannot be imposed from the outside.[16] Innovation driven by threat is the most powerful form, whether prompted by past defeat, as for the Reichswehr in the 1930s, or by looming challenge as Britain faced in the same period. The Defence Strategic Update 2020 highlights a number of increasing threats. These include a reduction in stability of the rules-based order, increased coercive activity in the region, accelerating military modernisation of other Indo-Pacific players, more assertive major powers, and the proliferation of emerging and disruptive technologies. By explicitly highlighting these threats, the Strategic Update offers reasons for change which can empower visionary leaders and provide the spur to increased innovation.

Trust in RAS is a particular challenge, given low risk tolerance and how heuristic biases distort assessments of probability. Humans have developed an expectation that machines will operate flawlessly; human nature is such that machine failure is not tolerated. A system that is able to not only understand but also act requires a new interpretation of that expectation. For example, there have been five fatalities involving autonomous cars worldwide since 2016 and consequently the adoption of this technology has slowed markedly. Compare this to the fact that in Australia in 2020 there have been over 730 fatalities on the roads with human drivers;[17] yet humans continue to operate cars. This highlights the relative acceptance that humans have of human judgements. Addressing the issue of trust of AI, autonomy and learning machines is a crucial undertaking. The recent issues around the Boeing 737 Max anti-stall functionality also highlight why automation must be approached cautiously. Trust is key to enabling to the realisation of RAS technology both in Defence and more widely.

The force design impacts and concepts of operations changes are naturally key areas for exploring the role of RAS. This was the topic of Army’s Future Land Warfare Branch experimentation program in 2018. This examination, including modelling with the Defence Science and Technology Group Joint Operational Analysis Division, focused on the value of RAS at the unit level. The experiment showed clear potential, not only for combat capabilities but also in the logistic and enabling elements. What was missing, however, was the next steps, the ‘so what’ and the ‘now what’ from both joint and single-service perspectives. Furthermore, the experiments tended to focus on RAS replacing humans in a like-for-like role. If RAS is approached purely through a ‘like today but better’ methodology then an important opportunity will be missed to perhaps do things differently. How this new technology informs our concept of joint warfighting in the future, and thus capability investment, is a key outcome. Force Design Division is currently authoring the ADF RAS Concept, which, while pre-decisional, is nevertheless being completed without an overall future joint warfighting concept. Force Design Division has also issued its Future Joint Logistic Concept. The absence of a guiding future joint operational concept means that, at best, subordinate documents represent exquisite shelf wear or, worse, simply a waste of effort. Both RAS and logistics are means rather than ends in their own right and should be approached as such, supporting the overarching concept.

In addressing how Defence seizes the opportunity provided by RAS, there is a need for clear guidance from the Joint Force Authority on a RAS electronic architecture. This is key to enable ADF-wide integrated performance of RAS systems that capability managers will develop within their domains. This is needed as a matter of urgency, and preferably should be aligned and coherent with our key security partners’ systems so that interoperability is assured by design. This would nest within the overarching vision of future joint warfighting, which is already using such integrated baselines.

A willingness to innovate is also stymied by process. The capability life cycle is optimised for the deliberate—read slow—procurement of large major systems. It is incompatible with the model of ownership of most technology in 2020: in the vernacular, a throwaway society. While self-evidently, Defence must remain a responsible government department and expend its budget effectively and with the utmost integrity, there is risk that norms of iterative caution will stymie prototyping and experimentation. In turn, capability opportunity or transient advantage may be lost.

The British Army is approaching the procurement problem through a notion it refers to as ‘Prototype Warfare’, defined ‘as a new approach to routine military activity that seeks to mimic the pace and intensity of wartime transformation by prioritising experimentation and adaptation to rapidly inform doctrine and practice’.[18] The Chief of the General Staff, General Sir Mark Carleton-Smith, has often mentioned that he wants to ‘invert the pyramid and empower the most junior in the Army to lend intellectual energy into the debate as to how warfare is changing in the Information Age’.[19] We must also take the opportunity to innovate and rapidly. It is heartening that Army is heading in this direction with the creation of the RAS Implementation and Coordination Office (RICO). The RICO has been undertaking exploration activities and prototyping to help Army understand and codify its RAS needs and wants and to gain insights to inform its future designs and user requirements. Navy has also initiated ongoing RAS exploration activities, including Autonomous Warrior in 2018.

Conclusion

This essay highlights that there are high payoff opportunities for RAS in the realm of joint logistics. It argues that the ADF might reimagine how it overcomes the challenges of both time and volume. One promising element is a vastly enhanced logistic decision-making system which exploits automation and artificial intelligence to distribute only what is needed, quickly and with precision. The system would deliver through the use of novel air vehicles and autonomous ground transport, which would dramatically improve efficiency and redundancy. However, these opportunities might be stymied by cultural challenges, both inside and external to the organisation. Feeding cultural resistance are real issues of trust in and of RAS, and, while the adoption of new technology has always presented challenges, some of the unique problems associated with RAS can be expected to require significant effort to overcome. Clearly articulated direction is needed to reinforce the intent already expressed by the Secretary of the Department of Defence and the Chief of the Defence Force. Force design and future concepts need to guide how RAS technology nests in an overall vision of how the joint force will fight and win in the future. The model should enable a capability life cycle that has the agility and flexibility needed to seize transient advantage. Finally, the joint enterprise needs a clear willingness to invest in prototyping and experiment in order to identify and seize the game-changing opportunities that RAS offers.

The future masters of technology will have to be light-hearted and intelligent. The machine easily masters the grim and the dumb.

Marshall McLuhan