The US is set to begin tests of an air-to-air combat drone, pushing the frontiers of artificial intelligence and human judgment in aerial warfare while attempting to ensure affordable mass for great power competition in the skies.
This month, Breaking Defense reported that General Atomics is set to begin flight tests in December of its design for the Defense Advanced Research Projects Agency (DARPA) LongShot program, which aims to develop an unmanned aircraft vehicle (UAV) that can launch air-to-air missiles.
General Atomics, a US energy and defense corporation headquartered in San Diego, was awarded a contract from DARPA for Phase 3 of its LongShot effort, which could be worth up to US$94 million, Breaking Defense reported.
The report notes that the drone is expected to significantly increase the engagement range and mission effectiveness of current 4th-generation jet fighters and air-to-air missiles, with flight testing validating primary vehicle handling characteristics and laying the foundation for follow-on development and testing.
Breaking Defense says that the flight testing will validate DARPA’s views on LongShot, a turducken-like unmanned aircraft system to be dropped from a bomber or fighter and potentially used by both the Air Force and Navy.
The same report notes that the project aims to develop a novel UAV that can significantly extend engagement ranges, increase mission effectiveness and reduce the risk to manned aircraft.
Breaking Defense notes that the Pentagon is prioritizing innovative drone design following the announcement by US Deputy Secretary of Defense Kathleen Hicks of the “Replicator” effort to acquire thousands of drones across multiple domains in the next 18-24 months.
Asia Times has reported about the Replicator project, which in addition to Hicks’ description aims to serve as a resilient distributed system offering quicker deployment closer to the tactical edge of operations.
Replicator also emphasizes ethics and compliance with armed conflict laws to mitigate concerns about autonomous systems in combat alongside multiple US autonomous drone initiatives aimed at revolutionizing future warfare capabilities.
Asia Times has also reported on the US Autonomous Multi-Domain Adaptive Swarms-of-Swarms (AMASS) project, which aims to develop autonomous drone swarms capable of being launched from sea, air and land to overwhelm enemy air defenses.
The project aims to create a system that can deploy and control many unmanned drones to neutralize an adversary’s defenses such as anti-aircraft systems, artillery, missile launchers and intelligence-gathering platforms. These drone swarms will primarily aim to deter or counter a potential Chinese invasion of Taiwan.
The LongShot drone may become an air-to-air loyal wingman. Asia Times has noted that developing loyal wingman drones addresses the need for mass-produced and disposable aircraft in a potential conflict between great powers.
These drones provide an advantage in numbers to their operators by serving as decoys, a swarm of drones or as a complement to manned aircraft, thereby enhancing their combat capabilities by extending the latter’s sensor ranges and by operating in areas deemed too dangerous for manned aircraft due to advanced surface-to-air defenses or other aerial threats.
While contemporary air combat and the LongShot drone may focus on missile-based beyond-visual-range (BVR) combat, the increasing stealth of both drones and fighter aircraft could mean that opposing forces might have difficulty finding each other, inadvertently ending up in a close-quarters dogfight. A dogfight is one of the most challenging aspects of air-to-air combat but recent AI experiments have shown promising results.
DARPA’s Air Combat Evolution showed that AI fighters scored cannon kills against human pilots due to their superhuman accuracy in aiming cannons from seemingly impossible angles, outperforming human pilots in classical, close-range, turning dogfights.
Underscoring those advantages, Michael Byrnes notes in a 2014 article for Air & Space Journal that AI’s superhuman level of machine-controlled maneuvering and accuracy makes every cannon round a “golden bullet,” meaning a single AI-powered unmanned fighter with just a few rounds of ammunition and sufficient fuel can wipe out an entire manned fighter fleet at marginal costs.
Byrnes points out that such costs can be further reduced by adopting commercial-off-the-shelf (COTS) hardware miniaturization of components and eliminating the need for a one-to-one crew-to-aircraft ratio.
Despite that, human pilots still have some significant – and possibly irreplaceable – advantages despite that overwhelming potential advantage.
In a May 2022 article for Engineering & Technology, Rich Wordsworth notes that the human eye’s ability to scan for threats continuously, passively and instantly focus is deceptively tricky to replicate.
He says that cameras don’t provide peripheral vision simultaneously as they can see what’s directly in front of them and that drones have often not been approved for flight in normal airspace due to the question of who’s looking out the front.
According to Wordsworth, conventional air combat is becoming increasingly expensive. While AI can reduce the risks to pilots and the limits of G-forces, it does not save human costs.
Wordsworth points out that remotely piloted aircraft are more expensive, as they require a large crew with different teams taking over from each other and significant support infrastructure.
Although AI technology has proven effective in certain situations, such as ones with clear-cut solutions, air combat’s complex and ever-changing nature means that human judgment will still be necessary to make high-risk decisions.
Despite the rapid advancements in AI capabilities, technology will for the foreseeable future only partially replace the need for human decision-making in dynamic air combat situations.
As such, the ideal man-machine framework for future air operations would operate on a paradigm that combines the adaptability of human judgment with the precision of AI automation.