Data as the Driver
Within the ML structure exists the concept of deep learning – using data to drive real-time decision-making.
“Deep learning is a subset of machine learning that develops a hierarchy of concepts using multilayer neural networks to extract higher-level features from raw input data,” said Dr. Derek Walvoord, Senior Manager, Image Science, L3Harris.
The most critical component in deep learning is obtaining and understanding the data itself. As L3Harris continues its path to digital engineering, the business is currently developing a digital data strategy, critical to leveraging possible AI solutions.
“L3Harris was recently entrusted by the U.S. Air Force to provide two of the three components for their AI platform, highlighting our expertise in data sets,” Walvoord said. “To support defense missions and reduce the burden of training data, L3Harris provides a trusted source of labeled synthetic training data to feed algorithms, considerate of intelligence and defense communities security.”
This fills a critical demand for advanced algorithm development and simultaneously empowers an army of digital analysts.
Customer needs continue to evolve as there is an increased focus on multi-domain operations. Information and decision superiority is the key to maximize operational effectiveness.
“AI/ML driven analytics will support workflows across all domains, determining information gaps and tasking appropriate sources to close said gaps,” Walvoord said.
These course of action analytics are what will enable automated effects in multi-domain environments and operations, maximizing efficiency for the modern-day warfighter.
Connecting Sensor to Shooter
L3Harris has a long history of developing sensors and remote sensing applications that provide critical data to customers worldwide. AI/ML-driven analytics help convert the sensor data to higher-level information, such as detections, and then to sense-making products that provide more meaningful representations of the data for exploitation by the warfighter.
The concept of operations (CONOPS) has traditionally been to plan, task, collect and process a mission’s sensor data from centralized data centers. Tasking is transmitted to the sensor platform; the collected remote sensing data is transmitted back to the data center; and the resulting products are then transmitted out to the warfighter in the field.
This centralization increases operation time and is fully dependent on the availability of communication chains.
“Pushing AI/ML processing of the sensor data to the edge devices – where the sensors are located – allows for an enhanced CONOPS where select operationally relevant products can be produced on-board and transmitted directly to the localized units that need it in much less time,” said Mike McGonagle, Senior Manager, Software Engineering, L3Harris.
The communications environment of today is extremely congested and contested. It is impacted by a variety of issues, including the ever-growing complexity of systems, terrain, malfunctions, and active and passive threats. Modern-day communication and sensor systems must overcome these threats via optimized tools that can monitor the environment and make sense of the data to meet real-time mission objectives without direct intervention.
“Our AI/ML technology fully transforms warfare from a reliance on a plan that is created from the ground in advance of a mission to providing real-time and learned adaptive knowledge directly to the teams that need it,” McGonagle said. “This transformation improves situational awareness by allowing the system to focus on intent rather than identification, enabling the transition from memorization to learning and adaptive behavior that can be rapidly incorporated and shared.”
AI-Enabled Autonomy
AI also elevates autonomic technologies. Acting as a force multiplier, AI solutions rapidly transform vast amounts of data into salient information for autonomous operations and for tactical aids. These aids are used for navigation and maneuvering both within the battlespace and also across the spectrum.
“Our maritime surface autonomy technology, as seen in our ASView Control System, allows us to detect, track and assess ships’ intent, supporting real-time decisions required for autonomous driving and collision avoidance,” said Dr. Marc Olivieri, Senior Fellow, L3Harris.
L3Harris’ proprietary Autonomous Surface Vehicle (ASV) control system not only optimizes autonomous and remote control of unmanned vehicles, it also enables the conversion of manned vessels for unmanned use.
“AI-driven decision aids are targeted today to accelerate our decision loop,” Olivieri said. “This November, our AINetAntx demonstration of Shadow Stalker in the NAVY’s Amazon Web Services (AWS) environment showcased the possibilities of AI-driven decision aid solutions."
“These can feed autonomic technologies for maneuvering through the spectrum and across the battle space to maximize survivability and the performance of effects. However, this is not just about defense applications, as our autonomy technology currently supports environmental protection efforts and safety of emergency response personnel in oil spill response,” he continued.
L3Harris is also researching net-centric collaborative autonomy, which will enable coordinated course of action (COA) engines. These COA solutions reinforce resilient networks and coordinate distributed operations of unmanned platforms/sensors-effectors pairings.
“Ultimately, these AI autonomic solutions will support man-unmanned teaming – managing disparate resources and meeting mission objectives using our Mission Intent Drivers ‘MIND’ technology and deep reasoning solutions for explainable and trusted AI,” Olivieri said.
Growth in Research
L3Harris is committed to the future of AI and ML capabilities not only with direct investment in internal research and development, but also within the industry overall. The company currently partners with several universities dedicated to AI and ML projects. It also supports the National Science Foundation Center for Big Learning, dedicated to exploring and pioneering research in emerging deep learning for data applications.
“Our AI Tech Network is comprised of L3Harris subject matter experts, and this dedicated team shares lessons learned and advocates for core capability investments in the space,” Brower said.
L3Harris is already leading in the development of AI technologies, and the impact AI will have on society in the next decade alone is undeniable.
“We are excited to embrace these transformational capabilities and outpace the technology of the past,” Brower said.
