A Keynote Address at I/ITSEC 2022
At the 2022 Interservice/Industry Training, Simulation and Education Conference (I/ITSEC), hosted by the National Training & Simulation Association (NTSA), Trideum Corporation CEO, Van Sullivan, presented a Keynote address to the audience challenging the I/ITSEC community to develop “Adaptive Decision Support Systems” —systems that adapt to both the situational data and the individual to improve critical decision making by leveraging data captured from all phases of the acquisition lifecycle, with a particular focus on training individuals and teams to improve critical decision making.
During the opening ceremonies “Fireside Chat,” featuring General David Allvin, Vice Chief of Staff of the U.S. Air Force, and Major General Shawn Bratton, Commander, Space Training and Readiness Command, both military leaders spoke of the need for coupling strategy with implementation through innovative education, training, doctrine, and test activities, and the need to accelerate the process to give our warfighters a timely advantage.
Mr. Sullivan responded to that call to action in his keynote address. This article is an abstract of the address.
“This year’s theme, Accelerate Change by Transforming Training – It’s Time to ACTT!!, is specifically designed to challenge us as professionals in modeling and simulation, training, and education to enhance, adapt, and accelerate our solutions to meet training and education requirements through advanced technology and approaches. Faced with declining resources, increased competition, and the ability to rapidly implement new technology across the community, we must be able to achieve transformation before the solutions and approaches themselves become obsolete. Let’s ACTT now to revolutionize training that will meet the challenges we face, both now and in the future. “
—Matt Spruill, 2022 I/ITSEC Conference Chair and Director, Training
Services & Solutions Sector for Trideum Corporation
Developing Adaptive Decision Support Systems
By Van Sullivan
In order to revolutionize training in a “declining resources” environment, we need to use leverage. Leverage is what happens when we use existing/emerging technology that other people are already developing to solve a problem. The convergence of testing and training will continue to produce opportunities for the test community to leverage capabilities developed by the training community and vice-versa.
Let’s tie that together with data and decision making. We need to work together as an industry– from requirements, system development, testing, and training, to sustainment—to leverage capabilities across the enterprise to create adaptive, personalized systems that enable more timely and higher quality decision making. I’m suggesting we harness this great brain trust within the I/ITSEC community and focus it not only on technical and tactical proficiency training, but also on training critical decision making; collecting data from training exercises and using that data to improve the design and functionality of both the tactical systems and the operators.
There are three parts to this challenge:
- Real-time, data collection and analysis with a situational assessment of the individual or team’s judgement and ability to make decisions with a set of standards and persistence
- Adaptive training to focus on areas for improvement, including selecting the optimal combination of information display, decision making, and system interaction. My new car does this – I configure the displays, audio system, seat and mirror adjustments, and save it to key 1. After my wife gets in and rearranges everything, all I have to do is start the car with Key 1 in my pocket and the car resets to my configuration. The result is a “coupled system” – the car and Van are coupled for an optimal experience.
- Leveraging the data and analysis from a training program and closely feeding that into the design of the system development, training, and sustainment process to accelerate change resulting in improved mission performance, more quickly.
I’ll illustrate all three and then tie it together with a system example.
Real-Time Data Collection and Analysis
For Part 1, real-time data collection and analysis, there are several methodologies with associated tests, data, and metrics to measure individual and group judgement and the ability to make decisions given a specific time and situation. I’ve used, and have been subjected to, several judgment/decision assessments and I’m amazed at how accurate and insightful they can be.
Regarding the Adaptive Decision Support System application, the data collection and analysis must be done in “human real-time” during training so that the analysis can impact the delivery of the training and the configuration of the system. As decision-making abilities change due to circumstances and environments, the data must include relevant information to describe the condition of the operator (tired, hangry, stressed) and the environment (noisy, hot, physically unstable) to provide the proper context of the data. Data models must be standardized, and the data must be shared on a persistent platform to enable leverage and reuse.
The data from those assessments can be used to align the better decision-makers with the more critical decisions. I’m proposing we ALSO use that information to develop an individualized, adaptive training program, to target specific areas to improve the decision-making abilities. Not only for individuals but also for teams.
That leads us into Part 2, Adaptive training to focus on improving critical decision making, including selecting the best combination of information display, decision making, and system interaction. Those are key components of an Adaptive Decision Support System. Adaptive as the system is configured to adapt to the unique person operating the system and assess their current inherent ability to make the decisions needed to execute the mission based on metrics and quantitative data and focuses the training to improve critical decision-making skills.
Instead of “manning the machine”, let’s “machine the man”– using a machine to aid the person / team to more efficiently and effectively accomplish their mission. By training the machine to be proficient at executing the “mundane” tasks like ingesting lots of data, then sorting and organizing that data into useful information with predefined priorities, the human can focus on the “complex” tasks – which typically involve decision-making where lives are at stake.
There are efforts to develop systems that ingest great amounts of data, identify data or trends that are significant, and alert the operator with actionable information—using User Experience/User centered design combined with artificial intelligence, machine learning, and simulation to increase both the technical and tactical proficiency of the warfighter. I’m suggesting we add decision-making proficiency to this paradigm.
- Use the training environment to objectively collect data in real-time to assess performance based on a human’s state, the system configuration, and the mission context.
- Use real-time analytics to provide instantaneous feedback on both the inherent ability of the person to make a decision, and the quality of the decision based on a “predictive” impact of that decision on the mission using simulation – just like we do in war-gaming exercises.
Now—how can we then leverage the information learned from the training system to improve the tactical systems?
Sharing and Leveraging Training Data
The third and final piece is sharing the information from the training environment and leveraging that data into the system development, training, and sustainment process. If we can assess the ability of individuals and teams to conduct a mission, measuring their technical, tactical, and decision-making proficiency and discover the “best tuned” configuration, we need to use that data to inform the system design, development, training and sustainment process. Of course, we need to have a standard data model and persistent data with access to users across the acquisition life cycle.
An Example of Machine Learning in Action
Let me give you a relevant example to illustrate the concept: a UAS ground control station with an embedded training system. We have extended the tactical system design to include some AI to process data to produce information (instead of just saying there is an unlit cell tower at this location, it determines whether that tower is an obstacle we need to avoid and provide an alert). We also have Machine Learning in the embedded training system to learn how the operator “best” uses the system and reconfigures both the displays and how the operator interacts with the system based on performance optimization. This is similar to my car example, but in this case the embedded training system “learns” how to best configure the tactical system based on mission performance, the specific operator, and the overall conditions (of both the operator and the environment.)
This user-specific data is saved and used to couple the system configuration with the operator to provide a fine-tuned pair for the mission.
Using that data collected from a broader set of operators, we can provide continuous feedback from the training environment to the system developers to the system maintainers to improve the overall performance of the coupled operator-system—a system that is “fine-tuned” to the operator, and how he or she ingests information to make decisions. As part of the sustainment program, the data related to specific operators and configurations are stored in a cloud so when Van logs onto the ground control station in a box in Tampa the system configures itself to best suit Van. It also assesses Van’s current decision-making abilities and “fine tunes” how he interacts with the system.
That’s a data intensive operation, with lots of user-centered design and implementation, tailored to the individual, or team. The training is actually used to refine the development and operation of the system in a constant feedback loop, giving the training community the opportunity to get further left of the bang and further accelerate capability change.
Different communities are already working pieces of this puzzle. Our challenge is to work together to leverage each other’s capabilities to accelerate change. I encourage you to find opportunities to collaborate and leverage the big brains in this IITSEC community.
Implementing Adaptive Decision Support Systems
Let’s use adaptive training to improve both the decision-making capabilities of our troops and the systems we field and align those decision-making abilities to the roles and responsibilities of the job.
Most of the capabilities required to design and implement Adaptive Decision Support Systems already exist. We don’t need to do basic research or violate the laws of physics, we need to agree to cooperate, agree to standards, identify the existing capabilities, integrate them into a usable system and then share the data.