Swarms at War: Chinese Advances in Swarm Intelligence

Source Link

By: Elsa Kania

The Chinese People’s Liberation Army (PLA) anticipates that future warfare will be “unmanned, invisible, and silent” (“无人、无形、无声”战争), with ever higher degrees of “intelligentization” (智能化). [1] PLA strategists expect that future, autonomous combat involving unmanned systems, as well as the joint operations of unmanned and manned systems, will have a dramatic impact on traditional operational models (PLA Daily, January 5, 2016). Future UAV swarming (无人机集群) will involve “intelligentized” and semi- or fully autonomous systems. [2] The PLA recognizes the disruptive potential of these techniques, which could be used for saturation assaults (饱和攻击) to overwhelm the defenses of high-value targets, including perhaps U.S. fighter jets or aircraft carriers (Science and Technology Daily, March 29; China News Network, November 2, 2016; China Military Online, December 31, 2016).

Chinese advances in artificial intelligence, including deep learning techniques, have enabled considerable progress in swarm intelligence. There is technical and conceptual research, development, and testing ongoing across Chinese academic institutions, the private sector, defense industry, and military research institutes to support such capabilities. At this point, given the limited information available and the relative opacity of these efforts, it is difficult to compare U.S. and Chinese advances in swarm intelligence. Nonetheless, the PLA has closely tracked U.S. initiatives focused on swarm tactics (e.g., Science and Technology Daily, March 29; Science and Technology Daily, May 30, 2016) and seeks to develop countermeasures and comparable capabilities. Looking forward, the PLA’s advances in intelligent unmanned systems and swarm tactics could serve as a force multiplier for its future military capabilities.

Chinese Breakthroughs in Swarm Intelligence

During the fall 2016 Zhuhai Airshow, official media prominently featured Chinese breakthroughs in swarm intelligence. To date, the efforts of the China Electronics Technology Group Corporation (CETC) appear to be the most advanced. In November 2016, CETC, in partnership with Tsinghua University and Posong Technology (泊松技术), revealed its progress in swarm intelligence with a formation of sixty-seven small fixed-wing UAVs utilizing autonomous swarm control and dynamic centerless networks with communication and coordination among UAVs (CETC, November 1, 2016; China Military Online, November 6, 2016). Such swarms could be used for reconnaissance, strike, jamming, and other missions. For instance, CGI sequence available in media reports at the time showed the swarm formation in action, first hunting and then dive-bombing and destroying an enemy missile launcher (Weibo, November 6, 2016). This technique possesses advantages in efficiency and survivability due to the distribution of capabilities across the system, as well as lower costs for offense relative to the difficulty of defending against a swarm.

In the spring of 2017, a formation of 1,000 UAVs at the Guangzhou Airshow by a private company reportedly again broke records (Global Times, February 14). At the time, military experts quoted in Chinese media similarly highlighted that this technique could be used to create a distributed system with payload modules mounted on small drones (Global Times, February 14). However, the actual sophistication of this particular effort—and potential linkage to military efforts—is unclear.

Once again, in June 2017, CETC demonstrated its advances in swarm intelligence with the test of 119 fixed-wing UAVs, beating its previous record of sixty-seven (Xinhua, June 11). This swarm engaged in catapult-assisted takeoffs and demonstrated complex formations. At the time, CETC commentary highlighted that swarm intelligence is future of intelligent unmanned systems, and CETC UAV expert Zhao Yanjie (赵彦杰) characterized future intelligent swarms as a disruptive force to “change rules of the game” in warfare (Xinhua, June 11).

Although it is difficult to compare the sophistication of U.S. and Chinese efforts, these high-profile Chinese tests have seemingly deliberately followed U.S. tests and demonstrations with larger ones of their own.

China’s National Initiatives in Artificial Intelligence

China’s future capabilities in swarm intelligence, military-use AI, will be enabled by high-level plans and extensive funding. China’s new national roadmap for artificial intelligence advances an ambitious agenda for the development of this critical emerging technology through 2030 (MoST, February 16). According to Minister of Science and Technology Wan Gang, this framework will focus on building up national capabilities in artificial intelligence, the applications of AI technologies, policies to handle resulting risks (e.g. job losses), and international collaboration (Xinhua, June 29). This development plan will also establish a special fund for research and development, while seeking to “rapidly gather AI talent,” including through encouraging foreign companies to establish R&D centers for AI technologies in China (South China Morning Post, June 29). This initiative, also referred to as “Artificial Intelligence 2.0,” will focus on areas including big data, intelligent sensing, cognitive computing, machine learning, and swarm intelligence (Xinhua, November 18).

Through China’s national strategy of military-civil fusion (军民融合), the PLA will seek to take advantage of relevant civilian advances in artificial intelligence to enable military applications. [3] At the CMC level, the PLA has reportedly established an Intelligent Unmanned Systems and Systems of Systems Science and Technology Domain Expert Group (军委智能无人系统及体系科学技术领域专家组) (Hunan Daily via Weibo, October 30, 2016). Such a group could be responsible for establishing strategic objectives and requirements, while perhaps also liaising with academia and industry. For instance, members of the group reportedly visited a new testing area for self-driving cars, which was also to be used to test military unmanned combat vehicles.

In practice, military-civil fusion could be achieved through partnerships between military and civilian research institutes or companies. For instance, in late 2016, the Military-Civil Fusion Intelligent Equipment Research Institute (军民融合智能装备研究院) was established as a collaboration between the North China University of Technology (北方工业大学) and a private technology company, Zhongbo Longhui (Beijing) Information Technology Company, Ltd. (中博龙辉(北京)信息技术股份有限公司) (China Science and Technology Online, November 28, 2016). The institute has received support from the Naval Equipment Research Institute, the Army Equipment Department, the Rocket Force’s Unit 96658, and other military organizations (People’s Daily, November 4, 2016). It will pursue research in intelligent robotics, artificial intelligence, unmanned systems, and military brain science.

Chinese national-level research and development efforts will likely enable civilian and military applications. In 2017, China established its first national deep learning laboratory. This lab, under the leadership of Baidu, one of China’s premier technology companies, in partnership with Tsinghua University, Beihang University, and the Chinese Academy of Sciences will research deep learning, computer vision and sensing, computer listening, biometric identification, and new forms of human-computer interaction (South China Morning Post, February 21, 2017). Of note, Beihang University is closely linked to the development of military aeronautical and astronautic technologies, including research focused on UAV swarming and manned-unmanned teaming (see below).

Future Research and Development

Although the PLA’s research on a number of topics and projects remains relatively nascent, it will be important to continue to track the progression of such research and development based on the available sources. The Central Military Commission (CMC) Equipment Development Department’s Scientific Research and Procurements Bureau (科研订购局) issued guidelines for pre-research funding under the 13th Five-Year Plan (Equipment Development Department, August 1, 2016), through which included funding will be directed to a number of topics that relate to or could enable UAV swarming, including: 

research on “bee swarm” (蜂群) UAVs on self-organizing network architectures, associated monitoring and control technologies, swarm networking and positioning technology, and network anti-jamming technologies. 

highly reliable autonomous flight control technology for new energy ultra-long endurance UAVs technologies for intelligent identification of targets and adaptive patterns for analysis based on deep reinforcement learning with large-scale remote sensing data methods for brain-like learning algorithms able to engage in sensing in unstructured environments 

As of 2017, the National Defense Science and Technology Key Laboratory Fund (国防科技重点实验室基金) has also committed to fund multiple projects related to artificial intelligence and UAV swarming, including the following: 

intelligent task planning technology to improve the management of unmanned swarms based on deep learning (China Aerospace Radio and Electronics Research Institute) 

artificial intelligence methods for unmanned vehicles to adapt in complex maritime environments (Harbin Engineering University) 

comprehensive decision-making, management, and control technology for advanced UAVs, including control and management technology for manned-unmanned cooperation, in order to achieve the integration, intelligentization, and networking of advanced aircraft (Beihang University) 

At present, multiple military and civilian research institutes appear to be working on swarming UAVs, based on their published research and patents on the topic, also including, but not limited to: [4] 

China Electronics Technology Group Corporation 

China Aerospace Science and Industry Corporation’s (CASIC) Third Institute’s UAV Technology Research Institute 

Harbin Institute of Technology’s National Key Laboratory of Robotic Systems and Engineering 

Tsinghua University 

Beihang University 

Harbin Engineering University 

Northwestern Polytechnical University 

The Disruptive Potential of Swarms at War

Looking forward, the PLA is clearly seeking the capability to leverage adaptive, intelligent unmanned systems across multiple domains of warfare, including with swarming tactics and manned-unmanned teaming. [5] The Academy of Military Science’s authoritative textbook, The Science of Military Strategy, anticipates future unmanned systems that utilize intelligent, nano, and micro technologies will have an “increasingly prominent function” on future land, sea, aerial, and space battlefields. [6] In the future, highly automated (自动化) and intelligent weaponry composed of unmanned systems is also anticipated to replace traditional weapons.

Indeed, PLA thinkers expect that future land, sea, air, and space battlefields will be full of unmanned combat weapons, forming a ‘multi-dimensional, multi-domain unmanned combat weapons battlefield system of systems’ (PLA Daily, February 21). The PLA has closely, consistently tracked U.S. efforts in swarm tactics through DARPA within the framework of the Third Offset strategy and related defense innovation initiatives (e.g., PLA Daily, May 18; Sohu, April 22). Although its approach is thus informed by U.S. efforts, the PLA’s future initiatives could diverge from that of the U.S., including with a focus on asymmetric applications that target high-value U.S. weapons platforms.

The PLA recognizes the disruptive operational potential of intelligentized unmanned systems and swarming tactics. Chinese advances in swarm intelligence enable this disruptive technology, considered “a breakthrough for future unmanned combat,” according to Zhao Jie (赵杰), director of the 863 Plan Intelligent Robotics Expert Group (China Military Online, November 6, 2016). In particular, the anticipated advantages of intelligent swarming UAVs include their functional distribution, high system survivability, and low operational cost, as CETC UAV export Zhao Yanjie has also noted (Sohu, April 22). These systems could engage in intelligence, surveillance, and reconnaissance (ISR), offensive operations, whether independently or in coordination with other weapons systems, as well as electronic warfare, thus supporting critical areas of strategic deterrence, operational confrontation, and tactical operations (Sohu, April 22).

China’s ‘Counter Offset’? – Competing to Innovate

Chinese advances in swarm intelligence—and the known research and development undertaken to date—constitute a critical example of its strategic competition with the U.S. in defense innovation. For the PLA, certain disruptive technologies, including those associated with the Third Offset, are believed to be “strategic commanding heights” (制高点) in great power competition, with the dual advantage of “effective deterrence” and victory in war-fighting (PLA Daily, May 7, 2016). For instance, future PLA intelligent unmanned systems could also serve as an asymmetric means through which to target high-value U.S. weapons systems, including aircraft carriers.

The PLA’s strategic objective of strengthening the military through science and technology (科技强军) will take advantage of a coordinated national strategy of “innovation-driven development.” The PLA seeks to “overtake [the U.S. military] around a corner” (弯道超车) through cutting ahead, rather than taking the same track, by achieving technological, conceptual, and organizational innovation in strategic frontier (战略前沿) technologies. Ongoing U.S. defense innovation initiatives must take into account the trajectory of Chinese military innovation in critical technological domains.

Elsa Kania is an analyst focused on the PLA’s strategic thinking on and advances in emerging technologies, including unmanned systems, artificial intelligence, and quantum technologies. Elsa is also in the process of co-founding a start-up research venture. Her professional experience has included working at the Department of Defense, FireEye, Inc., the Long Term Strategy Group, Harvard’s Belfer Center for Science and International Affairs, and the Carnegie-Tsinghua Center for Global Policy. She is fluent in Mandarin Chinese.

CB_17_9-1.pdf