AI in Security - Governmental and Commercial

of six

Autonomous Vehicles & Drones - Emerging Uses

The United States government is currently considering a wide variety of applications for Artificial Intelligence (AI). Currently, different government departments, particularly the Department of Defense (DOD), is leveraging AI to address operational issues. The key emerging applications or use cases in AI for autonomous vehicles and drones specific to the US government include logistics and autonomous military vehicle and drone development.

AI Use in Military Logistics

AI-Powered Autonomous Vehicles and Drones

of six

Autonomous Vehicles & Drones - Implementing Use Cases

The use of autonomous vehicles and drones has increased rapidly in the past few years, and the available technologies and their range of applications has broadened considerably. The military uses these technologies for a variety of applications, and heavy military investment in this sector has led to a proliferation of research and development. This report details some emerging uses for these technologies and the ways in which different companies and organizations are expanding their research, development, and production efforts to fit new and increasing demands.

Aircraft Maintenance

  • AI and drone technology are increasingly being used for aircraft maintenance in both defense and commercial applications.
  • Drones are useful in aircraft maintenance applications including: visual inspection of aircraft for potential lightning damage, inspection of aircraft for problem areas and scheduled maintenance, delivery of parts and tools to repair workers, and the automation of defect detection in potentially damaged aircraft.
  • The Air Force is using real-time sensor data from F-35s to develop predictive maintenance and repair systems and schedules for aircraft engines and other systems.
  • IBM's AI system Watson is being used by the US Army's Logistics Support Activity (LOGSA) to develop customized maintenance schedules for their Stryker fleet. The Watson AI system uses data pulled from 17 different sensors on each unit to predict and determine maintenance needs. Watson will also be used to analyze shipping and supply flows for repair parts and tools in order to determine the most cost and time efficient methods for the procurement of new supplies and parts.
  • Blue Bear Systems Research is a company that specializes in the development of autonomous and unmanned systems used in the defense, commercial, and emergency response sectors. Blue Bear has developed autonomous systems for aircraft inspection such as Smart Boomerang and RAPID (Remote Automated Plane Inspection and Dissemination). RAPID (created through a collaboration of Blue Bear and Output 42) uses high definition 3D imaging systems to collect and analyze inspection data and can reduce time required for aircraft inspection by up to 90%.
  • A wide variety of AI and robotics systems are being developed for different uses and aspects of MRO (Maintenance, Repair, and Overhaul). These include Invert Robotics crawling inspection robot, Lufthansa Technik's Mobile Robot for Fuselage Inspection (MORFI), Rolls Royce's development of robots for inspecting and servicing engine and machine parts that are difficult to access, and many more.

Network and Fleet Control

  • Control of coordinated fleets or swarms of drones is being researched for military applications, including the potential for F-35 pilots to control groups of drones flying near the cockpit. Control of coordinated fleets of drones or other unmanned vehicles is best achieved through integrated network systems.
  • Embedded Module technology developed by Persistent Systems is advertised as "the most advanced, most scalable, and most efficient Mobile Ad Hoc Networking (MANET) system in the world."
  • The Embedded Module can be used to combine unmanned aerial vehicles (UAVs), unmanned ground vehicles (UGVs), and sensors into a single network, and also contains an HD video encoder and Android computer system. Integration into a single network and the inclusion of video and computer systems make the Embedded Module one of the lightest, cheapest, most powerful and most efficient systems for operation UAVs and UGVs.
  • Since the Embedded Module integrates all components of unmanned vehicle operation into a single network, services from any integrated part of the network, such as video or sensor data, can be accessed from any node in the network from any location. The Embedded Module can also integrate disparate operating systems from different manufacturer's into one cohesive network, allowing unmanned vehicles and sensors from different manufacturers to easily and efficiently communicate.
  • Creating a common network for unmanned systems also allows for greater cohesion in control of larger fleets which increases the capacity for performing coordinated maneuvers such as swarming autonomy and collaborative behaviors. Feedback from the Android system of the Embedded Module allows for the use of swarming algorithms in unmanned systems and the synthesis of data from sensors to better coordinate swarm formations and avoid collisions.
  • The enhanced onboard data synthesis and processing enabled by the Embedded Module also drives better machine-made decisions without human input, allowing for more autonomous vehicle operation.
  • The US Army has contracted Persistent Systems in its increased use of drones and autonomous vehicles, and employs the Embedded Module system in its UGVs as part of its Common Robotic System-Individual (CSR(I)) program.

Target Identification

  • AI technology is being used in drones and other autonomous vehicles to improve and enable autonomous target identification and recognition.
  • The US Department of Defense is currently researching and developing technology that can be integrated into a UAV to allow the system to automatically Detect, Recognize, Classify, Identify (DRCI) and target specific personnel, locations, or other objects of interest. The Department of Defense is hoping to develop targeting technology that will implement machine learning algorithms such as Artificial Neural Networks that will enable increased flexibility and adjustability of target identification criteria and DRCI taxonomy to allow for broad application of the technology in varied environments.
  • The development of technologies with Automatic Target Recognition (ATR) and other DRCI functions will also require further research and development of sensor technology, communication systems, more advanced and specific algorithms for DRCI, and Human-Machine Interface (HMI).
  • In 2019 the US Air Force Research Lab and General Atomics demonstrated their Agile Condor project which uses AI and machine learning technology to allow MQ-9 Drones to autonomously identify and transmit the locations of targets over a large surveillance area. Agile Condor's ability to identify and transmit target locations without human input also greatly reduces the amount of satellite bandwidth required for its operation, thus increasing the efficiency of the program and decreasing its burden on communication and information systems.


  • AI is being used in defense and commercial applications to increase the autonomy of unmanned vehicles and systems.
  • One of the main areas of research for semiautonomous and autonomous vehicles is the development of AI programs and technologies that can be used for mapping, navigation, object classification and tracking, and obstacle avoidance.
  • Examples of development in this area of technology include the SLAM program created by Clearpath Robotics. SLAM (Simultaneous Localization and Mapping) is a program that uses innovative sensor integration and point cloud processing to construct a map of a robot's surroundings, and track the robot's own position in relation to that map. SLAM also incorporates object identification and tracking which allows the robot to identify and avoid obstacles, and tag points of interest in its environment. Clearpath also offers software technology packages like the Autonomy Research Kit which allows a robot to autonomously navigate an environment.
  • Thermal imaging is also an important component being utilized in the autonomization of vehicles. Thermal imaging devices such as the FLIR ADK are critical for Advanced Driver Automation Systems for pedestrian recognition and navigation of cluttered environments.
  • FLIR also makes a variety of other products and systems that increase performance of autonomous or unmanned vehicles such as the R80D drone which uses AI processing for object detection and classification and to allow semi-autonomous and autonomous operation.

Drone Disruption

  • As the use of drones and other autonomous vehicles in military operations becomes more common, there is an increased need and desire for counter-drone technology, or C-UAS (Counter-Unmanned Aircraft Systems) technology as well. C-UAS technologies are systems that are used to detect, intercept, or disrupt drone operation.
  • While the US Department of Defense has invested heavily in UAVs and AI, it has also increased its investment in C-UAS as the use of drones by ISIS and other groups in surveillance and attack operations has become more prevalent.
  • In 2018 there were 235 C-UAS products available from 155 different manufacturers. Of the 235, 88 were capable of detection only, 80 were capable of interdiction only, and 67 were capable of both detection and interdiction. This is a huge increase since 2015 when a survey by Sandia National Laboratories identified only 10 C-UAS products on the market.
  • So far the most common use of C-UAS has been in defending and protecting military bases, most often as a complement to other existing defense systems such as counter-mortar systems and surveillance systems, but there is growing interest in more portable C-UAS technology that could be used to protect mobile convoys and ground units.
  • Along with improvements in size and portability, C-UAS developers are also working to develop systems with more precise detection, identification, and classification capabilities, and with increased ability to fend off large numbers of drones at once.
  • The US Air Force has recently awarded a $23 million contract to Ascent Vision Technologies for mobile counter-drone vehicles which will employ "radar, optics, radio frequency detection, a jammer and electronic command and control mitigation to find and remove unknown UAS."
  • The US Army is also seeking to bolster its anti-drone defenses with anti-drone ray guns that will use radio control frequencies to disrupt drone communication systems and force drones out of the air.
  • Other C-UAS technologies still in development include mobile high-enery laser systems that can detect, track, and destroy UAV's; electromagnetic weapons that use short pulses of microwaves to incapacitate drones' electrical systems; and more advanced C-UAS technologies that will be fully autonomous and powered by AI.

Research Strategy

Findings from the Autonomous Vehicles & Drones — Emerging Uses report were used to inform the basis of this report and to identify emerging needs and focuses within this industry. Once an emerging need or use was identified, the research team browsed relevant news articles, indexes of companies who have contracts with the US military, company websites, and industry reports to identify how different companies and organizations are exploiting and addressing the emerging opportunities and uses for this kind of technology.
of six

Autonomous Vehicles & Drones - Investors and Researchers

Data Collective, Andreessen Horowitz, and Founder Collective are three VCs investing in seed and early-stage drone startups.

Data Collective, DCVC

  • Data Collective DCVC's website can be found here.


  • Data Collective is a San Francisco-based VC that funds "deep tech" entrepreneurs across all stages of the business cycle. The firm invests in artificial intelligence, big data, machine learning, cybersecurity, enterprise software, and health care ventures.
  • The key sectors the company invests in are industrial transformation, computational bio & health, geospatial intelligence, computational risk assessment, resiliency, and foundational technology.
  • It has existed for 20 years and is currently invested in over 150 companies. Data Collective's team comprises 35 experienced technology experts including Stanford & Berkley professors, CTOs, and chief scientists.
  • In the period August 2013 to August 2018, it made two seed investments in drone companies (including non-security).


  • AWS, Google Cloud, Microsoft Azure, VMware, and Illumina--companies that enabled large-scale AI and ML--are some early investments made by the company.
  • Data Collective, DCVC was the lead investor in the seed funding and Series A round of Fortem Technologies, an "autonomous AI-enabled interceptor drone" manufacturer.
  • It has also invested in Evolv Technology, a company that uses AI and sensor technology for its security scan devices that are used by governmental agencies and private businesses. The VC has also invested in Embark, an autonomous truck company.
  • Data Collective also made a seed-level investment in Drone Deploy, "a cloud-based drone mapping and analytics platform".

Andreessen Horowitz

  • Andreessen Horowitz's website can be found here.


  • Andreessen Horowitz is a Silicon Valley-based VC that was founded in 2009 and backs tech entrepreneurs. It has $2.7 billion in assets under management and invests in seed, early stages, and late stages.
  • The firm is led by individuals who are former CEOs, CTOs, and founders of technology companies with a wide range of collective technical expertise ranging from security, biotechnology, to crypto.
  • Andreessen Horowitz invests in "consumer, enterprise, bio/healthcare, crypto, and fintech".
  • In the period August 2013 to August 2018, it made two seed investments and two early-stage investments in drone companies (including non-security).


  • Andreessen Horowitz made seed-level and Series A investments in SykSafe, a company that manufactures interceptor drones for airspace security, in 2016 and 2017, respectively.
  • The company has made early-stage investments in Zipline, a medical products-delivery drone company, and Shield AI, a firm that builds AI systems for "intelligence, surveillance, and reconnaissance (ISR) defense operations".
  • It also made seed and Series A investments in Skydio, a company that claims to make the "most advanced flying AI on the planet".

Founder Collective

  • Founder Collective's website can be found here.


  • Founder Collective is a seed-stage investor based in Cambridge, Massachusetts. It has been in existence since 2009.
  • It invests in advertising, e-commerce, health care, hardware, internet of things, mobile, marketplace, SaaS, and software.
  • The partners at the firm have all founded technology companies. They invest in companies that they believe are building "something important" and "live and die" by their seed investments.


  • Founder Collective has invested in SkySafe, a company that manufactures interceptor drones for airspace security; Shield AI, a firm that builds AI systems for "intelligence, surveillance, and reconnaissance (ISR) defense operations", and Hivemapper, a company that makes "mapping, visualization, and analytic tools" for machines such as drones.

Research Strategy

The VCs that most actively (slide 30) made seed-and-early-stages investments in drone companies between August 2013 and August 2018 were Draper Associates, SV Angel, Founder Collective, Social Capital, Andreessen Horowitz, 3one4 Capital, Data Collective, Homebrew, Lemnos, Commercial Drone Fund, Qualcomm Ventures, New Enterprise Associates, Sony Innovation Fund, Airbus Ventures, Lux Capital, Accel Partners, General Catalyst Partners, Sequoia Capital, and GV. Among these investors, we identified three (there are more) that have made investments in companies manufacturing drones for the governmental or commercial security sector in the seed and early stages.

of six

Machine Learning/Modeling (Data) - Emerging Uses

AI-powered machine learning for security intelligence and machine learning in surveillance and reconnaissance for the security personnel are some emerging use cases in the AI for machine learning in the governmental/commercial security sector. Project Maven uses AI and machine learning to provide security intelligence from videos captured on drones. It does this by identifying hostile activities and persons of interest to national security. Hivemind Nova from Shield AI provides the security personnel with surveillance and reconnaissance capabilities during urgent or unfavorable situations. It does this through machine learning which helps the robot to operate autonomously as well as teach other robots and collaborate with them to complete the mission with minimal human intervention.

AI-powered Machine Learning for Security Intelligence

  • Due to the large amounts of data readily available, AI is set to be critical in security intelligence. As stated repeatedly by experts and top-ranking security personnel in the U.S., it is imperative to incorporate AI and machine learning into security intelligence.
  • Project Maven is a Pentagon initiative designed to use artificial intelligence and machine learning to differentiate objects and people in lengthy drone video footage. It can pinpoint images of high interest to the security department from moving or still images. The first phase entails intelligence processing automation to streamline the identification of hostile activities.
  • The project started in 2017 with Google as the contract holder to provide AI services and expertise. Google's AI tool known as TensorFlow was used to design the program and train the algorithm. After one year, Google failed to renew the contract when its employees raised ethical concerns over the company's involvement with the Pentagon.
  • Soon after, Plantir took over Google's role in the project. The company powers CPB and ICE's surveillance networks and is also involved with creating software for the police.
  • Maven's intended goal is to automate the job done by human analysts to sift through videos looking for insights and actionable data. This will free their time which will be directed towards making informed decisions based on insights delivered by Maven.

Machine Learning in Surveillance and Reconnaissance for Security Personnel

  • The use of automated tools in the security sector has been in development for a long time but the combination of AI and machine learning has gained traction recently.
  • Hivemind Nova from Shield AI is a quadcopter drone designed to assist security personnel in surveillance and clearing operations. The drone is powered by Hivemind which is a machine learning application that makes it possible for machines "to see and learn about the world."
  • This system allows the security personnel to access urban areas, building interiors, tunnels, caves, high threat environments and areas with no GPS coverage to study the level of potential threats. To operate the system, users can command it from a Hivemind-enabled mobile unit and the company says that a pilot is not needed and the amount of training is minimal.
  • Once the drone is airborne, it maps the target area while transmitting a video to the mobile unit. The machine learning technology helps the system to learn from the environment without human intervention hence lowering the risk to security personnel. It can also teach other robots about the target area and work with them to accomplish the mission within a shorter time.
  • The Hivemind Nova is intended to protect the security personnel and civilians by using AI and machine learning. The aim is to provide actionable insights during urgent situations and limiting spaces and terrain.

Research Strategy

To provide the emerging applications/use cases in AI for machine learning within the governmental or commercial sector, we consulted the web for reports, case studies, press releases, and industry experts' discussions about the subject. We managed to find reports and other sources with details about the emerging use cases in AI and machine learning. We limited our selection to the applications that directly touch on the government/commercial security sector.

of six

Machine Learning/Modeling (Data) - Implementing Use Cases

AI-powered machine learning in the security sector is becoming a critical piece to security and surveillance. Companies utilize and benefit from machine learning in many ways. AI has produced security concerns, such as video analytics, detailed surveillance of consumers, and potentially dangerous government projects. However, it has also provided for many security advantages, including fraud and malware detection, the prevention of cybersecurity threats, and the evolution of machine learning and protection over time.

AI-powered machine learning

  • AI-powered machine learning has become a critical piece in security surveillance. In fact, about 42% of countries worldwide are utilizing and exploiting machine learning technology for surveillance purposes.
  • Use of machine learning is not particularly limited to the security sector; there are many other entities that exploit machine learning. Examples include business that use AI to determine and improve upon analytics, traffic monitoring by city officials, and the surveillance of citizens either lawfully or unlawfully.
  • According to "The Malicious Use of Articial Intelligence" machine learning and AI could be deleterious if put to use in digital, physical, and political facets.

Dangerous exploitation of AI-powered surveillance

  • The American Civil Liberties (ACLA) released a report warning citizens that current AI advancement may allow for end users to pick up on unusual behaviors (kissing, hugging, general dispositions, etc.).
  • Jay Stanley, senior policy analyst at ACLA stated: " We face the prospect of an army of AI security guards being placed behind those lenses that are actually, in a meaningful way, monitoring us, making decisions about us, scrutinizing us."
  • Although there have been tens of millions of devices that could be used by AI for machine learning surveillance for many years, constant monitoring has only recently become plausible for the public.
  • Video analytics monitored by businesses are now changing that. Self-driving cars, retailers, and a wide range of business all stand to benefit from this sort of constant surveillance. However, this does pose a huge consequential privacy concern for consumers.
  • Project Maven is an initiative designed by the Pentagon to develop algorithms and integrated video data to assist military personnel. Google was originally working with the Pentagon on this project until ethical concerns were brought up. Google employees were concerned that their own AI could be weaponized and stated that Google "should not be in the business of war."
  • The Pentagon itself plans to exploit machine learning in Phase 2 of this project, and aims to create "actionable intelligence."

Beneficial uses of AI and machine learning

  • Many companies are instead choosing to benefit from potential positive outcomes of AI, such as fraud and malware detection, network interruptions, and intrusion detection. Whether the machine learning is monitored or lacking any human supervision, AI algorithms have been proven to be effective in detecting fraud online and reducing accessibility from potential threats.
  • In fact, 80% of executives in the telecommunications sector believe that AI is critical to preventing cyber attacks in their organization. Additionally, 51% of companies rely on AI for threat detection and response.
  • As cyber threats become more advanced over time, companies are utilizing AI to detect unknown threats proactively, analyze a large amount of data that would otherwise be humanly impossible, and protect infrastructures from possible data breaches.
  • Additionally, machine learning is being utilized by many companies to not only act proactively but to use artificial analysis to learn and evolve over time, improving processes along the way.

Research Strategy

We began this research by searching sources that discussed machine learning prevalence in general. We wanted to take a look at credible sources that ascertained whether AI had expanded around the world. Then, we specifically searched for sources relative to AI in the United States, along with an understanding of whether its presence in the security industry is pertinent.

We then began taking a look at companies or government agencies that had used AI and machine learning for a specific, purposeful goal. We reviewed a report that summarized the findings of 26 researchers regarding potential threats surrounding AI in the security sector. A comparison was between the aforementioned theme of potentially dangerous initiatives about AI with several other credible sources. We focused on both the benefits and threats that machine learning provides for as it relates to security and surveillance.
of six

Machine Learning/Modeling (Data) - Investors and Researchers

Charles River Ventures (CRV) and Dell Technologies Capital (DTC) are some early stage investors in AI/ML specific to the commercial security sector.

Charles River Ventures (CRV)

How the company is involved

Dell Technologies Capital (DTC)

  • Dell Technologies Capital, founded in 2012, is an early-stage venture firm that has backed about 81 startups, spanning "a broad range of technology areas."
  • The technology areas the firm focuses on include but are not limited to "the Internet of Things (IoT), AI, ML, developer-led Infrastructure, next-gen infrastructure, server-less computing, and security."
  • Startups within the Dell Technologies Capital portfolio are currently experiencing 100% growth rates, and several exceed $50 million in revenue.

How the company is involved

  • Dell Technologies Capital has invested in three firms (Jask, Zscaler and RedLock) that offer artificial intelligence for commercial security.
  • Jask, a company delivering a solution that uses artificial intelligence to help security operations centers monitor threats in real-time, received about $12 million in Series A funding from DTC.
  • DTC has also backed RedLock, a company that offers a platform that uses AI to "monitor Microsoft Azure, Google Cloud Platform and Amazon Web Services (AWS) for cloud security and compliance risks," with "$8.5 million in a Series A funding round."

Dive deeper

Only the project owner can select the next research path.
Need related research? Let's launch your next project!

From Part 02
  • "Introducing the most advanced, most scalable, and most efficient Mobile Ad Hoc Networking (MANET) system in the world, now in a SWaP-timized form factor: the Embedded Module."
From Part 05