Two U.S.-based robotics companies, Boston Dynamics and Ghost Robotics, produce quadruped unmanned ground vehicles, or robot dogs. At last year’s Association of the U.S. Army’s annual conference, Ghost Robotics unveiled an armed version of the “dog,” equipped with a 6.5mm Creedmoor assault rifle, as part of a partnership with Sword International.

The technology, which typically costs $50,000 to $150,000 depending on the company and model, requires further development before it can head to the battlefield, according to officials with the Army’s robotics program.

“These legged platforms have some promises which we’ve identified, primarily from a mobility standpoint,” Milot Resyli, the branch chief for dismounted robotic systems at the Army’s Ground Vehicle Systems Center, told C4ISRNET. “There are limitations to them as well from an endurance [perspective], as well as the payload capability and power of how much they can support.”

The development of a robot dog has been almost two decades in the making. In 2004, the Defense Advanced Research Projects Agency launched the BigDog project in hopes the robot could serve as a pack mule and accompany soldiers in terrain too rough for conventional vehicles.

Both the Air Force and the Department of Homeland Security have explored using the platforms for patrols. This summer, the Space Force conducted a demonstration at Cape Canaveral Space Force Station, Florida, to illustrate how ground-based robots can automate manual and repetitive tasks, such as patrols.

The Army has not yet fielded one, according to the service’s chief roboticist, Robert Sadowski. However, he told C4ISRNET, the Army does have pilot programs underway as well as its “10x Project,” which tries out technology to upgrade future infantry forces.

Roboticists with the 10x Project have experimented with the platforms to assess how they might be used and pushed to do more missions than the current level of technology allows, Resyli said.

But he also said the technology has room to grow in its ability to operate in all weather conditions and in shallow water.

Unlike costly, larger robotics platforms, the Army doesn’t anticipate keeping its smaller robots around for more than a decade. Instead, current programs of record for small systems will undergo a technology refresh at the 10-year mark.

According to Resyli, the next opportunity for legged platforms to compete for those programs is at the refresh point, which falls within the next five to 10 years. He added, however, that there are talks right now about the next iteration of these small platforms.

Unlike wheeled or tracked robots, which can fall over on rough terrain, legged robots can climb stairs and operate over rough surfaces. The quadruped platforms can also carry payloads, such as cameras and sensors, allowing them to act as early warning systems and put distance between the warfighter and threats.

“There are many other missions out there, specifically the subterranean missions where current track-and-wheel platforms are severely limited because you have potentially slight, extensive elevation changes — whether it’s stairs or steep slopes — that the current platform can’t support,” Resyli said.

Sadowski suggested the platforms could take the place of surveillance missions that would otherwise use a quadcopter, a type of flying drone that uses four rotors or propellers to fly.

Although quadcopters are used for a wide range of military and tactical surveillance missions, the technology struggles to support longer missions and heavier payloads due to the heavy weight of sensors and a limited power capacity — areas where quadruped platforms boast better capacities, Sadowski said.

For its part, the U.S. Army created the Joint Counter-Small Unmanned Aircraft Systems Office to address the proliferation of adversarial drones. The JCO came about after then-Defense Secretary Mark Esper in 2019 designated the Army as the executive agent for counter-UAS activities.

Maj. Gen. Sean Gainey serves as director of that office, and he spoke with C4ISRNET in September about what it takes to protect American interests against drones. This interview was edited for length and clarity.

The Army’s Fires Center of Excellence at Fort Sill, Oklahoma, focuses on counter-UAS capabilities. How is that effort going?

We’re going to transition what we’re doing at Yuma [Proving Ground in Arizona] into Fort Sill, and that process is happening right now. We’re starting to get the classrooms. They’re starting the program of instruction development. We’re working with the joint team across courses to make sure we’re maximizing all the [program of instruction] capability.

Will the center’s curriculum focus on different types of counter-UAS strategies, such as kinetic options versus electronic warfare options?

The curriculum will cover all of that. You’ll have an “operators’ course” that will really introduce operators to the spectrum of counter-UAS capabilities that are out there, really focus in on the systems the Army’s currently fielding at divisions and the other services are fielding at their sites or locations.

We’re really excited for our “planners’ course” because, right now, we really don’t have the planning or understanding of how to employ counter-UAS capability from a planner perspective, and that’s going to be a critical piece. That’s probably the course that’s going to be stood up first because, right now, as we’re putting capability out in our installation site, one of the biggest challenges is installation commanders having a planner that they can look at to do the employment of the capability and then lay out all the planning factors with it.

You recently talked about the need for more widespread knowledge across the force on counter-UAS systems. What is the JCO doing to facilitate that dissemination of training and knowledge across the entire force?

The first thing we did was put out a strategy. The strategy was really the first document that we put out that framed the problem for the entire enterprise. Then we put out a handbook based off what we’re doing forward in U.S. Central Command’s area of responsibility, capturing all the lessons learned. We not only provided that for the CENTCOM area of responsibility, but globally to all the combatant commands.

Right now, we’re on our fourth JKO [Joint Knowledge Online] module. If you go to JKO online, there are four modules there that span from operator-level familiarization all the way through policy implementations of employing counter-UAS capability. They’re actually really good modules, and we’re seeing a lot of interest out there.

Are those modules a requirement?

It hasn’t been updated as a requirement, but my assumptions are, as we continue to move forward and continue to mature, the counter-UAS training requirements will become a requirement in the future. Right now, it’s available, and we’ll find that with soldiers who are preparing for preemployment training, most are more likely to leverage it.

One of the most recent rounds of U.S. assistance for Ukraine included the Vampire system made by L3Harris Technologies. The Pentagon called it a counter-drone system, but L3Harris didn’t advertise it as such. What’s the working definition of counter-drone at the moment?

If you look at some of our capabilities that we’re having a lot of success with in theater now, they weren’t put over there for that purpose. We’ve essentially maximized the capability in that system to give it a dual role, whether it’s taking down rockets or mortars; they now can take down drones. That’s just a natural evolution in the capability because the threat is so prevalent and so large.

You can’t have just a c-UAS capability everywhere. You’ve got to be able to leverage whatever capability you have. So yes, I would tell you that anything that has the ability to take down a drone may not be classified [as c-UAS], but will be utilized in our system-layered approach.

What safeguards are used for autonomous technologies?

At least right now, the human-in-the-loop piece is going to be there until there’s continuing comfort in the ability of a system. But you can start building autonomy without actually having somebody push the button. You can get to the point where it detects, and then the operator does the final push, the effects piece. At some point, that’ll be a strategic decision: When do you want to fully autonomize the ability to let this system shoot? Especially in the future you have things like swarms and other types of threats coming in at you and need to be able to rapidly respond to some of the capabilities.

Some of the autonomy will more than likely have to be built in here, but [the question is]: How far do you go with the autonomy? I don’t think right now anyone is suggesting that we walk away from the system and just let the system do it itself from a defeat perspective, but we’re still looking to leverage technology that can get us there. Eventually, we’ll start having those policy discussions on how we move forward with the autonomous defeat.

How have lessons from the war in Ukraine shaped — or are currently impacting — decisions within your command?

What we’re seeing is what we’ve expected. We’re seeing that the EW capability we’re using works well. We see the kinetic ability working well. What it’s reemphasized is the threat’s ability to leverage drones and UASs to their advantage. If you don’t have a robust capability inside of your force, integrated air and missile defense system, a layered integrated approach with the common [command and control], you’re going to have challenges against these threats; and having it all the way down to the operator level to be able to respond at the operator level.

We’re seeing all of those types of trends employed and reinforcing what we need in order to ensure we’re doing [the right thing] as we continue to move forward and shape our forces.

Although autonomous vehicles have slowly made their way into the commercial realm, the Pentagon’s efforts to build robotic vehicles have been almost two decades in the making. For its part, the Army considers combat vehicles a top modernization priority.

Winners from the competition will deliver the first prototype units by 2028, according to Maj. Cory Wallace, the robotic requirements lead with the Army’s Next Generation Combat Vehicle Cross-Functional Team. In the meantime, programmers are taking a new look at how they will test the technology before it hits the battlefield, and that could include taking lessons from the civilian world.

“With the RCVs, we are looking at such new systems and new approaches,” Wallace said in an interview with C4ISRNET. “A lot of the testing procedures that are currently employed are not the most efficient.”

Military combat vehicles have historically needed to pass certain benchmarks, such as a specific number of test miles driven. The robotic vehicles, which are intended to be autonomous, might not need to pass the same thresholds, he said.

Instead, Wallace said, the service could look to certify the vehicles in a virtual environment as opposed to actually driving the vehicles the same number of miles required for regular combat vehicles. Testing and evaluation metrics might also look to companies already in the autonomous vehicle space to see what could be replicated in the military sphere.

Programmers have “been very open to try new means of testing,” Wallace said. “We’re trying to draft [off] what we’ve done in the civilian world, as well, to see if we can gain efficiencies by implementing their testing standards.”

Over the next five years, the Army plans to spend nearly $750 million on developing light robotic vehicles, according to fiscal 2023 budget justification documents. While the bulk of the funding will go toward developing the robotic platform and its control system, the Army will run a second acquisition program in parallel that focuses on software and building out autonomous capabilities that will eventually be added to the platform.

When testing the software capabilities, Wallace emphasized the need to develop a method that looks at changes between iterations rather than checking each line of code. If programmers get bogged down in evaluating the software’s code, he said, the program could lose momentum during the certification process.

By separating out the software acquisition pathway, the service aims to develop software-centric capabilities that can support any future robotic combat vehicle configuration, Maj. Gen. Glenn Dean, the Army’s program executive officer for ground combat systems, told C4ISRNET.

“A core element of the program that we’re executing is about establishing the base-vehicle architecture and the base-software architecture so that you can come in and subsequently add capabilities as modules of software that don’t require you to go back and test every other module of software to verify that it still functions appropriately,” he said.

The Army has sought to develop three types of robotic vehicles — light, medium and heavy — to use as scouts or escorts for crewed fighting vehicles, beginning with the light program.

Surrogate prototypes of both the light and medium versions underwent testing this summer at Fort Hood, Texas. However, in the spring, the service said it would first focus on smaller platforms and defer work on medium vehicles.

According to Brig. Gen. Geoffrey Norman, the director of the Next Generation Combat Vehicles Cross-Functional Team, the Fort Hood testing and future experiments help examine the range of capabilities that fit with a light vehicle versus a medium platform.

Once those experiments lead to refined requirements, the team will start on a specific program focused on medium-sized vehicles, Dean said.

The Robotic Autonomy in Complex Environments with Resiliency, or RACER, project aims to develop autonomous uncrewed ground vehicles to maneuver in unstructured off-road terrains. This round of testing began last week at Camp Roberts, located midway between Los Angeles and San Francisco, and will conclude on Sep. 27.

“One of the things that RACER is trying to do is push the speed, and we want to ultimately get to speeds that are faster than what combat vehicles are doing,” said Stuart Young, the project’s program manager, in an interview with C4ISRNET. “In order to do that…you have to see further out; you have to understand the world further out, and that’s particularly challenging in a hilly off-road environment.”

Researchers with the Defense Advanced Research Projects Agency previously tested the vehicles in a Mojave desert terrain this past spring at Fort Irwin. During the experiment, they focused on gauging the vehicles’ perception capability, testing whether the automobiles had the ability to differentiate between different types of terrain, such as grass, sand and rocks — all of which appear similar to a robot.

This time, DARPA is looking to see if the vehicles can plan ahead when confronted with objects like trees, hills and ditches that require the technology to see further away and understand the world farther out.

By giving the vehicles the ability to predict more about the world in front of them, they’ll be able to make the same calculations humans do about how to interact with — and go around — the objects in front of them.

Unlike many other vehicle programs underway at the Pentagon, Young said RACER isn’t focused on adding sensors. Researchers with the program use the same sensors they started out with when RACER began, choosing instead to rely on training machine learning approaches, such as neural networks, which mimic the human brain through a set of algorithms.

Robots have historically struggled with speed due to unpredictable environments that may feature obstacles or circumstances the autonomous technology wasn’t prepared to address, causing them to an error in their behavior or stop entirely.

What’s worse for robots, water or trees?

Contrary to popular thinking among the public, Young said the most problematic obstacle isn’t trees but rather water features and points where the ground drops off, or “negative obstacles.” The dipping points present a falling or tipping hazard for the robot, especially if the vehicle is traveling at high speeds.

“We need to be autonomous vehicles need to be out in front of the formation, not lagging behind,” Young said. “That’s the recognition that RACER had, which was robots are too slow, and they’re too brittle to be effective.”

The program ultimately wants to make its vehicles run at similar speeds as combat vehicles.

While autonomous cars made by Tesla and other commercial companies run at the same speeds as cars driven by people, usually up to 80 mph, combat vehicles, run much slower. The Army’s Bradley Fighting Vehicle, for example, typically has a maximum speed of between 30 and 40 mph depending on the model.

“We don’t have any structure in the environment that we can depend on, so we’re not going on roads, or even on trails,” Young said. “We’re literally going across the terrain.”

The RACER program runs through FY2025 and is divided into two phases, the first of which concludes next March. By the end of Phase I, researchers want the vehicles to reach just over 11 mph. At the end of the program, Young said they’re aiming for 19 mph.

The Camp Roberts experiment is number two of eight. While some future experiments will return back to the same terrains to compare measurements and to test improvements, others will take researchers to different biomes, such as terrains with more lush or swampy vegetation, Young said.

“In some environments, we’re doing better than we expected,” he said. “We’re pushing into environments where we still have lots of challenges and exploring those spaces earlier than I expected.”

“It’s true that most of my career has been in AI, but I don’t think most of my work here at DoD will be about AI,” Craig Martell, the Pentagon’s first Chief Digital and AI Officer, said at the 2022 Intelligence & National Security Summit in National Harbor, Maryland.

The Pentagon tapped Martell, a former head of machine learning at Lyft, to oversee its AI efforts this spring. The office subsumed multiple other defense digital authorities, including the Joint Artificial Intelligence Center and the Defense Digital Services, and reached full operational capability on June 1.

While much of the talk about the CDAO has focused on its role as the Pentagon’s leading AI authority, Martell said more attention and energy needs to be paid to data, which he said serves as the basis of the office’s “hierarchy of needs.”

In order to address the challenge of AI, the Pentagon needs to focus less on top-down AI decrees, he said, adding that while data decrees are still necessary, top-down AI restrictions hinder progress.

By focusing more on controlling data, the Pentagon can build a support structure that allows for heterogeneity among the different actors and companies working on AI projects, he said.

“It shouldn’t be centralized,” he said. “If we centralize it, we’re losing what I think is the biggest strength of the American industrial academic base.”

After building out data, Martell said the CDAO should focus on metrics, noting that outcome-based measurement, and not just effort-based measures, ought to be the standard. From metrics should come analytics, with AI coming in only once the needed infrastructure is in place.

“We have a very large AI challenge ahead of us but in order to do this, we have to get the data right,” he said.

Democrats and Republicans on the Small Business committees in both the House and Senate convened a so-called four corners meeting this week to negotiate draft legislation, which is in its final stages.

Rep. Blaine Luetkemeyer, R-Mo., the ranking member on the House Small Business Committee, said Thursday that the committees hope to finish drafting legislation that reauthorizes the Small Business Administration’s Innovation Research and Technology Transfer awards — which are made jointly with 11 federal agencies — by the end of the week.

“We’ve got a general agreement already with the four corners on this, but the devil’s always in the details,” he told Defense News. “We want to make sure that what comes out in text is what we agreed to in concept.”

“We’re knocking out the text as we speak, so everyone can see it today or tomorrow,” Luetkemeyer added. “Hopefully next week we can our side of the bill firmed it up and get it through.”

Both Rep. Nydia Velazquez, D-N.Y., and Sen. Ben Cardin, D-Md., who chair the Small Business Committee in their respective chambers, noted that negotiators have made “a lot of progress” on the compromise reauthorization.

Velazquez told Defense News that she feels “optimistic” and that she expects final legislation “soon.”

Sen. Rand Paul, R-Ky., the ranking member on the Senate Small Business Committee, downplayed the negotiations on Wednesday and told Defense News that they have not assuaged his concerns.

Paul has argued that the program lacks protections against ties between program awardees and China, and that some companies rely entirely on SBIR grants to sustain themselves without spinning off new businesses or products.

“Dr. Paul will not reauthorize this program without reforms to strengthen research security and stop abusive behavior by bad actors lining their pockets with taxpayer dollars at the expense of new small businesses with emerging technologies being able to access SBIR awards,” a spokesperson for the senator told Defense News in June.

Reauthorization of the program was originally included in an earlier version of the CHIPS and Science bill, but the provision was not included in the final version of the legislation that passed in July.

The total budget for the 40-year-old program ballooned to nearly $3.3 billion in 2019, with the Department of Defense accounting for the majority of the awards. Individual grants range from tens of thousands of dollars to more than a million for a two-year grant.

SBIR provides funding in three phases, with grant amounts typically increasing throughout each stage. For defense SBIR awards, companies are eligible to submit multiple project ideas, which some say leads to companies applying for multiple Phase I awards without ever maturing the proposed technology.

“We’re trying to find ways to incentivize and do the work yet provide the oversight to continue to do the work ― not just ripping off the program to continue to get these grants and at the end of the day they’re just making money, but they’re not actually producing products,” said Luetkemeyer.

“For some entities, for some different departments, there aren’t that many alternatives to go to…to be able to produce or develop a particular product or service for the Defense Department‚” he added. “We’ve got to be very careful that we don’t preclude the ability of companies to be able to meet our defense needs by hamstringing this bill.”

Contrasting standards for large, small companies

Kea Matory, director of legislative policy at the National Defense Industrial Association trade group, said in an interview that criticisms of awarding companies too many Phase I grants signal contrasting standards for small and large companies.

“Lots of our bigs get multiple awards,” she said. “We would never tell one of our large primes ‘oh, you’ve had too many, you need to sit down.’”

Matory added that delays in reauthorizing the program are already impacting small businesses.

“A lot of them are already hearing that SBIR could go away,” she said. “It’s like a game of telephone as it gets passed along to each person; it sounds more doomsday.”

When it comes to innovation, small businesses frequently bring new ideas to the table that could help the Pentagon with its modernization goals, she said. Without SBIR, many of those companies would be disincentivized to want to do business in the government and defense sectors.

The Pentagon will not award new SBIR grants starting Oct. 1 if the legislators fail to reauthorize the program. Ongoing contracts may continue but will not receive further SBIR/STTR funding.

If the draft legislation wins over Paul, the House and Senate could reauthorize SBIR as stand-alone legislation by the end of the month.

And if that doesn’t work, Senate Armed Services Committee Chairman Jack Reed, D-R.I., told the Defense News conference that negotiations are also underway to include SBIR reauthorization in the continuing resolution that Congress needs to pass by the end of the month in order to fund the government and avoid a shutdown.

“I would prefer that we get reauthorization as a stand-alone bill versus just a straight reauthorization where we stick it into a [continuing resolution],” said Luetkemeyer. “That really doesn’t solve problems. It extends the same problems that we have now.”

As a last resort, Congress could reauthorize the program in the Fiscal 2023 National Defense Authorization Act — although Congress is unlikely to finalize that legislation until after the SBIR/STTR grants expire.

The House NDAA, which passed 329-101 in July, has a provision that reauthorizes the program without any changes. The Senate must still pass its version of the bill before both chambers agree on final legislation — a process that usually takes several weeks.

“It is going to be the defining feature of global politics for the rest of our lives,” Bob Work, who served as deputy defense secretary in both the Obama and Trump administrations, told reporters Monday. “It is going to determine who is the greatest economic power in the 21st century. It’s going to determine who is the greatest military power. It is a competition that we simply must win.”

The nearly 200-page assessment, called the “Mid-Decade Challenges to National Competitiveness,” is the first published by the Special Competitive Studies Project, a private group led by Eric Schmidt, former Google CEO and co-chairman of the U.S government’s National Security Commission on Artificial Intelligence, and Work, who serves on the group’s board of advisors.

The organization seeks to build on the work completed by the congressionally mandated AI commission, which identified technology as the central element of the rivalry between the U.S. and China. The commission wrapped up its work last October.

According to the report, the years 2025 to 2030 will prove critical in deciding whether the U.S. keeps pace or falls behind in the technology battle.

Losing the competition could comprise Americans’ daily lives, the report said. Not only could China use its techno-economic advantage for political leverage, but Chinese domination could threaten free access to the internet and create a dependence on the country for most core digital technologies, making nations vulnerable to cyber attacks.

“Up to this point, because of the 20 years we spent in the Middle East, it kind of took our eyes off the ball,” Work said. “As this technological rivalry and competition was really growing in strength, we didn’t really respond as we normally have done in the past.”

Three technology battlegrounds — microelectronics, fifth-generation wireless technology (5G), and AI — tell the story of the U.S. and its allies coming perilously close to ceding the strategic technology landscape, the report said.

Those technologies represent the critical hardware, network infrastructure and software underpinning everyday life in the U.S. as well as the country’s national security apparatus.

Despite recent congressional action, the report’s authors said that progress in those fields illustrates reactive policymaking and a disconnect between the private and public sectors. More fundamentally, the report said the U.S. has failed to connect technology developments to strategic competition.

During the call with reporters, Schmidt specifically pointed to the danger posed by bioweapons designed using AI. Using AI, actors could treat biology and chemistry akin to building blocks, rapidly mixing and matching different combinations of compounds to create viruses.

“It’s going to be possible for bad actors to take the large databases of how biology works, and use it to generate things which hurt human beings,” he said.

The report also emphasized the likelihood of great power conflict, particularly involving Russia and China, in the years to come. While combat in traditional domains could still play a significant role, the report said that warfare will also be waged with and against industrial and financial power and pit innovation ecosystems against each other.

Schmidt, who spoke while coming back from a 36-hour visit to Ukraine, highlighted the role information technologies have played thus far in the country’s conflict with Russia. By moving the country’s data to the cloud, Ukrainian leaders were able to protect the country’s data from Russian hackers and strikes.

In future conflicts, the use of technologies will only expand, the report stated. To get ahead on these key technologies, the report called on the U.S. military to implement “Offset-X Strategy” to make the military “better prepared and positioned to outsmart, outpace, outmaneuver, and — as necessary — outgun the People’s Liberation Army.”

The approach focuses on expanding network-based operations and more dispersed, smaller units equipped with sensors and battle management software to make warfighters more situationally aware. The report also called for increased human-machine collaboration and an emphasis on software.

To make investments in these areas, the report asked Congress to include items in its next National Defense Authorization Bill that align with key technology areas and the “Offset-X Strategy.”

The components of the strategy did not contain any recommendations on lethal autonomous weapon systems. However, the organization wrote that any application of emerging technologies for military purposes can and should be done in ways that are consistent with the laws of armed conflict.

The company aims to create simulation models that run through and pinpoint different options for providing supplies and logistical support in the event of a national security situation to those in the decision-making room.

“If we can build that, then we can give planners a substantially different tool set to use for operational level planning and in its best incarnation, that set of tools also becomes available as decision support tools for the command and control units in the network,” said retired Air Force General Paul Selva, who serves as the company’s chief strategy officer, in an interview with C4ISRNET.

Selva served as vice chairman of the Joint Chiefs of Staff from 2015-2019 and prior to that as commander of U.S. Transportation Command.

DEFCON AI wants to use AI to comb through data on available options while also manipulating variables in a “simulation fabric” that replicates the real world in hypothetical environments, he said. The platform could then compare situations tested in the fake environment against existing real-world historical data on mobility and logistics operations.

After making a technical proposal to the Air Force, the company received a grant for over a million dollars to produce the software within 20 months, beginning July 2022.

“Fast-tracking innovative technologies is more important than ever in this new era of great power competition,” said former U.S. Secretary of Defense and DEFCON AI board member Mark Esper, in a statement. “If the U.S. military is going to modernize and maintain its overmatch against China and Russia in the years ahead, then there needs to be a much different and better relationship between the Pentagon and the most innovative sectors of our economy.”

When a major event with national security implications occurs, such as an airlift or a hurricane, organizing logistics is similar to controlling different cogs in a machine. The order in which tasks are completed plays a large role in the success of an operation and each part impacts the other moving pieces.

During Hurricane Katrina, for example, responding defense officials focused on establishing infrastructure over other priorities such as search and rescue or housing because the disaster destroyed nearby airfields. Without the runways, needed commodities including water, food and medical supplies struggled to reach the city.

The sequencing of actions during these types of events takes time because people need to go through data and look for relationships that work. DEFCON’s anticipated tool, Selva said, uses AI to complete the sorting while still giving end users the ability to test options via the platform’s “simulation fabric.”

“Our belief is AI alone doesn’t actually solve the problems that we’re being asked to solve,” he said. “It can’t be a black box that says just go do X.”

The idea for the company emerged after a client inquiry as to whether such a logistics simulation tool was available on the market, Selva said, adding that the only simulation product that came to mind was multi-player video games.

While much of the company’s oversight and management comes from the national security world, its technical talent pool draws from those who have worked in the software or gaming sectors.

Although the grant period is limited, Selva said DEFCON AI is planning to “move as fast as we can without breaking things.”

Autonomous vehicles remain a long-time goal, both in the commercial and military sectors. At a panel held by the Defense Advanced Research Projects Agency at Colorado State University on Wednesday, researchers and officials looked at the challenges standing in the way of vehicles capable of full autonomy in any environment.

“There’s a lot of technological barriers,” said Lt. Col. Chris Orlowski, who manages the Army office responsible for executing the hardware and software programs for robotic combat vehicles. He said some that are often overlooked include “our safety.”

This summer, the Army tested different types of robotic combat vehicles at Fort Hood, Texas. During the exercise, soldiers tested RCV platforms’ capacity by driving them on various terrain conditions commonly encountered on the battlefield. The tests also looked at the vehicles’ ability to identify and avoid obstacles and their capacity to fire weapons while moving.

Although the tests were considered a success, fielding truly autonomous vehicles is still years away. At the Fort Hood testing, a safety observer followed every platform in case of an incident during each exercise, according to Orlowski.

“We need to move away from that, especially when we want to field these provably safe systems in the future,” he said.

In recent years, autonomous vehicle ventures from companies such as Tesla have started to break into the commercial sector. This summer, regulators in San Francisco gave a robotic taxi service the green light earlier to begin charging passengers.

The rollout of robotic vehicles in a military capacity has been much slower. Although DARPA, the Pentagon’s research agency, has been working on self-driving vehicles for close to two decades, military robotics vehicles have to deal with a problem not found in the streets of San Francisco and other U.S. cities: the lack of roads and disarray expected on battlefields.

On regular streets, robotics vehicles have to account for expected obstacles like pedestrians, garbage cans and mailboxes. Meanwhile, the battlefield presents a whole host of different obstacles.

“You have to figure out how other things are moving and if they’re really adversarial, that makes it all the more difficult because maybe they want to move in a way that doesn’t make it easy for you to figure that out,” said Deva Ramanan, the director of the Carnegie Mellon University Argo AI Center for Autonomous Vehicle Research.

Outside of safety precautions, the Army also needs to examine its contracting process for robotic vehicles, Orlowski said. While it’s one thing to set up budgets, he said that it’s another to start leveraging authorities and signal to the defense industry that it’s time to start developing products.

The Army said this month that it is preparing to make future decisions on the potential acquisition and use of unmanned systems in combat by conducting additional soldier training and evaluation sessions with RCV platforms over the next 36 months.

When the vehicles are eventually brought to the battlefield, Orlowski said that he sees the technology being parceled out on an individual unit basis to augment the unit’s capabilities rather than putting them together in a group formation.

The Army lacks force structure to give every company, battalion and brigade electronic warfare capabilities or systems to counter drones. This gap could be filled by providing one or two autonomous vehicles with those capabilities to units, he said.

“I think where they offer promise is the ability to bring payloads that we can’t put on a manned platform today because we don’t have the size, weight power, capacity to do so,” Orlowski said.

While Thomas Jefferson and John Quincy Adams were fluent in several languages, Biden, like most U.S. presidents, is only known to speak English.

The video was a piece of synthetic media, more commonly known as a “deepfake.” Created using a combination of machine learning and artificial intelligence, deepfakes are hyperrealistic videos that replace one person’s likeness with that of another, or appear to show them doing something they never did.

And as the technology improves, they get harder to detect.

The workshop, which featured DeepMedia, a synthetic media start-up, focused on providing airmen with an introduction to deepfakes and their relationship to military applications as well as the company’s AI technologies that could be used to combat the threat they can pose.

Earlier this year, a deepfake video of Ukraine President Volodymyr Zelenskyy showed him calling on his soldiers to lay down their arms and surrender amid the country’s conflict with Russia. Although the video was quickly debunked and taken down, the technology is increasingly being used in propaganda and disinformation campaigns.

Rijul Gupta, a co-founder of the Oakland, California-based DeepMedia, called deepfakes the “next frontier” of video technologies, citing their low production cost and relatively quick turnaround time.

“Everything will be able to be faked in real-time,” he said at the workshop.

The company created the Biden video to illustrate that not all deepfakes are harmful. DeepMedia’s AI-based universal translator tool, for example, can be used to facilitate diplomatic communications, it said.

The translator tool takes video of someone speaking one language and transforms it so the subject appears and sounds to be speaking a different language in real time. The technology could also be used to pinpoint the linguistic benchmarks that differentiate a deepfake from a real video.

Hollywood has been using some form of deepfake technology for years, to make older actors appear decades younger of even provide new “life” to actors who have died. Actor Peter Cushing appeared in “Rogue One: A Star Wars Story” in 2016, more than two decades after his death.

Still, unauthorized deepfakes of celebrities and other politicians have spread widely across social media video platforms, prompting concern about the potential use of synthetic media to spread disinformation.

According to a 2021 report from the U.S. Department of Homeland Security, the threat of deepfakes does not come from the technology used to create them but from people’s natural inclination to believe what they see. As a result, deepfakes do not need to be particularly advanced or believable in order to be effective in spreading misinformation, the report said.

DeepMedia focuses on both detecting and producing synthetic media. In April, the Air Force Research Laboratory announced a partnership with the company to study the detection of deepfakes.

One of the main ways to detect whether a video is a deepfake is via an analysis of the language being used, which can reveal dialect differences between the video’s subject and audio presented.

Analyzing videos for those differences takes time, money and effort, said Emma Brown, a co-founder of the company.

In many cases, the cryptologic linguist conducting the analysis must have a deep understanding of the language and its dialects as well as the country’s culture and history. Simply having a cursory knowledge of the language isn’t enough to pinpoint the flaws in deepfakes, she said.

As the military struggles with a major recruiting crisis, attracting the needed talent to fill cryptologic linguist positions poses yet another challenge. To substitute human talent, Brown and Gupta explained that DeepMedia has developed AI tools that take the place of linguists.

“There’s simply not enough people to do this without AI,” Gupta said.