Training Human-Human and Human-Machine Teaming in the Digital Realm

With the increasing digitisation of the battlefield and greater automation, the crew of armoured fighting vehicles must be trained in ever more complex scenarios. MS&T Guest Writer Dr. Trevor Dobbins explores the role of simulation and the move from analogue to digital.

Operational success, particularly for small unit operations, has been shown to be based on the concept of Relative Superiority (RS), a concept developed in 1993 by Admiral William McRaven (who later, oversaw the mission to remove Osama Bin Laden in 2011). The concept is based on six principles: Simplicity, Security, Repetition, Surprise, Speed and Purpose. Of the six principles, two are highlighted here: Simplicity, the need for a simple plan and a robust set of Standard Operating Procedures (SOPs) to reduce operational risk in potentially complex situations, and; Repetition, practice – practice – practice, both the SOPs and mission rehearsal.

McRaven also highlights operational Single Points of Failure (SPF) where emphasis must be placed to ensure they are overcome, and points during an operation where RS, and therefore the likelihood of operational success is dramatically increased. Often these points are the same or coincidental and what operational planners need to focus on when developing systems and SOPs to reduce the risk from SPFs and increase RS.

Traditionally the training of SOPs and mission rehearsals has been undertaken in the real world. However, with the increasing capability of simulation and synthetic environments (SE), can they enhance training and rehearsal to increase operational effectiveness and potentially reduce risk, time and cost? Simulation has the potential to support trialling existing SOPs in new and novel environments/situations; developing and trialling new SOPs; repeatedly rehearsing the operation, including actions-on, with immediate After-Action Review (AAR). Because the operators will experience the operation during the mission rehearsal by being immersed in the simulation, it is possible to iteratively develop the operational plan to reduce risk (SPF) and enhance the likelihood of success (RS).

Simple SOPs are an essential part of ensuring RS within an operation. The repetition of SOPs in training, required for RS, is straightforward for learning the basics but their execution in the required range of conditions and environments is costly in terms of both time and money. Therefore, there is the opportunity for simulation to support this requirement providing unique training opportunities not available in the real world. Such opportunities include training to failure where the trainee can experience the ultimate result of their actions and mistakes can be played out to provide enhanced learning; replay the actions that resulted in failure to understand the outcome cause(s) and where operation success started to slip away; understand where Situation Awareness (SA) was starting to fail and so the wrong decisions began; and demonstrate that the individual’s/team’s focus was on the wrong cues.

Training Effectiveness & ROI

Does training simulation work? In some cases, this is difficult or impossible to prove, but there are examples where it is part of a recognised training programme and the qualification process, e.g. aviation pilot training and maritime bridge operations. But it also depends on the definition of training effectiveness. It is essential to consider the Return on Investment (ROI) of simulation including those not directly influencing the training outcome. For example, managing exposure to noise and vibration (complying with legal limits and reducing risk to health and injury); reduced platform/system maintenance costs; sustainability, and; easier/cheaper to update to new vehicles/platforms/systems.

As live training is costly, simulation has the potential to make it more valuable. Learn the basic SOPs in simulation (e.g. system functionality and communication protocols) so costly real training can focus on developing operational capabilities (e.g. Team SOPs) rather than learning the basics in an expensive-to-run vehicle.

How do we Make Decisions?  Situation Awareness (SA) is an inherent part of operational success. It is the basis of the decision-making process which occurs at the local, tactical and strategic levels: Local – the immediate surroundings in visual range, typically supported by camera systems; Tactical – the surroundings of direct influence, typically displayed on a map, and; Strategic – HQ functionality of the wider operation.  SA and the subsequent decision making requires the platform system(s) to provide the right information at the right time – therefore the system developers need to understand the operation, the CONOPS, and the decisions that have to be made. It must be understood that SA is a mental model of the world held with the individual’s head or between a team working together to complete a shared goal. SA is not achieved simply by pressing the button on a display. The configuration of the information on the display may even reduce SA by providing information in a way that results in conflicting mental models and confusion.

Coordinating and communicating between the crew of an Infantry Fighting Vehicle, here a British Army Warrior, and their dismounts can be a challenging team task. Image credit: Bohemia Interactive Simulations.

A simulated operation is able to force the individuals and teams to make mission critical decisions and therefore learn from making both right and wrong decisions and understand the optimum timing of decisions. Making an effective decision earlier means having the required SA and recognising the cues needed to support the decision making. SA is no longer developed by only the human operator; it is developed by the interaction of the human and the machine as part of a Joint Cognitive System (JCS). If done properly, the design of the system is optimised to make the most of the humans and the machines capabilities – the human typically doing the creative part and the machine acquiring and remembering the information and making it available when required.

Boots & Bots

But not only does a vehicle (e.g. AFV) Commander have access to camera systems and digital mapping with information overlays, they are increasingly having access to information from and control over off-board systems (e.g. UGV). But how do they control the UGV? Is it by remote control, does the UGV have automation for completing the Commanders control inputs, or is the UGV autonomous with the Commander able to make requests of it? Understanding our relationships with robotic and autonomous systems (RAS) is essential for future system development.

For the Commander/operator it does not matter if the UGV is real hardware in the field or a simulation, the sensor feed that they see and perceive is the same.  The military can hence potentially minimise their training with real UGVs, resulting in large cost savings. Similarly, simulation can be used to develop and test new CONOPS for joint manned and unmanned systems operations. Brutzman and Fitzpatrick (2020) investigated this opportunity and concluded “shared virtual environments can potentially be used during force-development efforts to plan for the integration of human-machine teams … it is quite clear that the use of live, virtual, constructive (LVC) simulations to wargame these capabilities becomes fundamental for all progress”.

This increasing focus on Human-Machine Teaming (MOD JCN 1/18) and Manned-Unmanned teaming (MUM-T), with the goal to optimise the system of systems, seeks to ensure the humans and machines do what they do best to maximise operational effectiveness/reduce risk by improving SA and subsequent decision making. How can simulation provide a better understanding and optimisation of MUM-T? It can for example reduce the operational risk at the SPF, and; bring forward the point where RS is increased, or increase the level of RS, engaging enemy forces before they become a risk to the human operators.  Within the simulation, human and human-machine teams can practice their response to unmanned entities.  In some ways this may be no different from confronting a human enemy. However, it is essential to develop effective response strategies and SOPs, particularly for responding to UXV swarms.

Where platforms/vehicles are relatively inexpensive in the relation to the simulator, the ROI calculation becomes more difficult. The simulator may cost more that the vehicle, therefore a good ROI might be gained by having embedded simulation in each vehicle.  Platform digitisation, e.g. via a Generic Vehicle Architecture (GVA), means that every platform can be a training system – simply by injecting a simulated view into the camera/sight feeds (Local SA) and supporting the mapping functionality (Tactical SA). This means that individual crew members can practice SOPs as well as sharing the simulation with other platforms to train the required interaction and distributed command and control. An example of this kind of system is the AJAX Appended Trainer where four vehicles are linked.

The Digital Progression

There are many examples around the world where large, impressive simulators sit quietly gathering dust, only to be activated for a VIP visit. Capital spend is relatively easy, whereas instigating an effective simulation-based training system is a challenge for most established organisations. It requires digital evangelists to generate and maintain the momentum required to get a digital training programme established; it is very easy to allow the analogue inertia to be maintained.

But training is not simple, there are many logistical and administrative hurdles to overcome, and supporting a complex simulator can itself be challenging. A training programme for 20 students may be primarily based on learning individual skills and then developing 10 effective team pairs, e.g. driver/navigator and commander/gunner. A traditional training approach might train the 10 pairs in parallel, but using a single expensive simulator means the pairs are trained in serial, drastically increasing training time and the trainers have to find activities for the nine pairs not using the simulator; an unacceptable result and poor ROI. Therefore, the overarching training concept must be optimised to account for the required skills and competencies against the costs and administrative/logistic constraints.

The change process is difficult – can this be a gradual iterative process, or does it need to be a big-bang? Example questions to achieve this include, ‘what training serials do we remove to replace with simulation as we can’t increase training time’? For some applications the best introduction is to make a dramatic change to digital, which means developing the simulation-based programme alongside the traditional programme and then switching over once it’s been validated and approved. But it will be rare for training to go exclusively digital, therefore the answer is an effective blend of live and virtual. Example digital-based training may include eLearning, desktop/PC simulation, part-task trainers, full-mission simulation, live with simulated feeds, embedded simulation training, data analytics, etc.

Train-The-Trainer is an essential part for the effective implementation of a digital training system. But there are an increasing number of ‘digital natives’ to support and facilitate the implementation. Future operators and trainers will have grown-up with computer games and digital media and will have no problem engaging with digital systems. The issue is the transition from those who have grown up with only live training to the digital natives; will the older instructors take the red pill?

The news proliferates with articles about the use of Big Data and its analysis. The digital simulation-based training system has the ability to generate vast amounts of data. The question is what data to collect and analyse? To answer these questions, it is essential to understand what good performance looks like, what metrics objectively describe what good is, and how these metrics can track training performance. Similarly, how can newly developed SOPs and their measures of performance be incorporated into the system and can they be common across all training scenarios? To maximise the effectiveness of the training system, the data collection and analysis needs to be automated so that the instructors can see the results in real time. This data aggregation also provides the ability to give the organisation leadership an immediate overview of their units training and the level of performance they have reached and maintained over time.

The digital battlefield is here, particularly with the increasing development and employment of network enabled systems. The proliferation of data has the potential to enhance operational effectiveness, but, the individual needs to know the location of the required information within the ocean of data being streamed to them – it is easy to drown in the ocean.  Understanding SA and the decision(s) to be made by the individual/team (including MUM-T) are essential to developing both the operational platform and the training system. An effective training system is more than the sum of its parts, but there is typically an 80-20 solution that can provide a training system with a good ROI.

Following this introduction to the topic of training in the digital realm, the next article develops the concepts surrounding the effective integration of simulators into training programmes, how they develop individual/crew performance, and what digital-first training can look like.

About the Author

Dr. Trevor Dobbins is the HFI Lead at RBSL (Rheinmetall BAE Systems Land) overseeing and developing the HF capability. He previously worked on land systems at Rheinmetall Defence UK and General Dynamics Land Systems, and on a range of projects at the DERA/QinetiQ Centre for Human Sciences, and as a Human Sciences Advisor for the MoD. He has authored many papers and reports including as the principal author of the High-Speed Craft Human Factors Engineering Design Guide.