The recent regional conflicts involving the Israeli "Egret" drone and the American "Titan" anti drone system have shown that identification of friend or foe has become a common issue faced by both drone and anti drone operations. For unmanned aerial vehicle (UAV) operations, in order to achieve precise target recognition, it is necessary to have powerful computer vision recognition capabilities, supplemented by massive target data annotation; For anti drone operations, anti drone systems not only need to detect targets in a timely manner, but also accurately identify target characteristics and friendly or enemy attributes. The traditional identification friend or foe system is mainly designed for manned aircraft and cannot effectively respond to various drone threats. Foreign media reported that in the past, when there were fewer drones, the problem of identifying friend or foe was not prominent. With the increasing number of drones on the battlefield, this problem has become increasingly difficult. According to foreign media reports, in this round of the Israeli Palestinian conflict, about 40% of the Israeli Defense Forces' drones were destroyed by their own personnel. Israeli Defense Forces soldiers will immediately destroy drones once they are detected during combat. The reason behind this is that in just a few seconds, from the discovery of the drone to the drone launching the attack, soldiers had no time to determine the identity of the drone. In February of this year, the German frigate "Hesse" carrying out combat escort missions in the Red Sea launched two "Standard" -2 anti-aircraft missiles at a target that could not be identified and was determined to be a "reconnaissance drone". Afterwards, it was proven that the targeted drone was a US MQ-9 "Reaper" drone that was performing a mission nearby. However, a spokesperson for the German Ministry of Defense emphasized that the "Hesse" had completed the full identification of friend or foe process, including radio response, before firing, and asked if there were any drones carrying out missions in the airspace. In fact, in actual conflicts, due to the inability to quickly identify friend or foe, soldiers from both sides of the conflict often conduct indiscriminate strikes on drones overhead to prevent bombs from suddenly falling on their heads. Currently, the main equipment used to discover and identify unmanned aerial vehicles (UAVs) is the radio scanner, which identifies issues and risks with the equipment. It can receive and recognize communication signals from drones, but it may also receive false signals and lack reliability. In addition, wireless scanners cannot identify the identity of drones, not only for individual use but also for many large anti drone air defense systems. So, is it feasible to equip drones with identity recognition devices? The answer is that the size, weight, and power consumption of existing identity recognition devices are too large to be installed on drones. For example, a micro identification device developed by a subsidiary of "Russian Electronics" holding company weighs only 150 grams and has low power consumption, but the accompanying airborne transponder weighs 60 kilograms and consumes over 660 watts of power, making it impossible to be carried by small and medium-sized drones. In fact, the weight of current airborne responders is difficult for most drones to afford. In March 2021, a US company launched an airborne transponder with friend or foe identification capability. This is currently the world's smallest certified airborne transponder. Compared with traditional airborne responders, its size, weight, and power are significantly reduced, with a weight of only one-third of the latter and a power consumption of one fourth of other airborne responders. It mainly provides identification of friend or foe capability for tactical unmanned aerial vehicle combat teams. It remains to be seen whether this world's smallest airborne transponder can solve the problems of excessive weight and volume. Even if the weight and volume issues of the airborne transponder are resolved, there are still information protection problems. Compared to large combat aircraft, the damage rate of unmanned aerial vehicles is very high. Once a drone equipped with this airborne transponder is shot down, the enemy will quickly obtain the transponder and crack its signal algorithm. Therefore, the practice of installing identification and response systems on drones is not accepted by most countries, which is also the main reason for the slow progress of this technology. The current unmanned aerial vehicles still use traditional photoelectric detection and recognition technology, and the development direction of this technology is to introduce artificial intelligence to solve the problem of battlefield target recognition. At present, the optoelectronic detection system used for drones has developed to the third generation, with white light/passive infrared imaging and laser ranging functions, and artificial intelligence technology is also being introduced. The first generation of artificial intelligence recognition technology was affected by computing speed and algorithms, resulting in a small number of recognized targets and the inability to identify men, women, or vehicle models, which easily led to misjudgments. The second-generation artificial intelligence recognition technology can automatically identify men, women, and vehicle models, and can also find targets in complex environments. The third-generation artificial intelligence recognition technology has a certain degree of self-learning ability and can find targets and launch attacks among hundreds or thousands of people. In some regional conflicts, both sides have used this technology to carry out precise strikes against specific personnel of the other side. For anti drone systems, it is also necessary to introduce artificial intelligence technology. For example, the US Army and Raytheon are accelerating the development of an anti drone radar that can automatically monitor and track small targets such as enemy drones, continuously track their flight trajectories, and transmit relevant information to ground commanders. Based on this information, the commander chooses to use high-energy microwave weapons or missiles to attack the target drone. Raytheon has been developing data fusion algorithms for this anti drone system, utilizing artificial intelligence technology to analyze the collected data and optimize key data. In addition, in recent years, the US Department of Defense has repeatedly purchased the "Titan" anti drone system, which is driven by artificial intelligence and machine learning technology, allowing operators to understand the surrounding environment within 5 minutes and provide protection on the battlefield. In addition, some unmanned aerial vehicle (UAV) detection and identification equipment currently under development in the United States and Russia can capture information such as UAV model, serial number, current location, and operator position, and can track multiple targets from a distance. Overall, the application of artificial intelligence technology in drone recognition devices is not widespread. For most drones on the battlefield, recognition is still achieved by comparing intercepted drone signals with signal features stored in scanners. However, the application of artificial intelligence technology in friend or foe identification systems will be the trend. Artificial intelligence can improve the automation, accuracy, and response speed of friend or foe identification systems. Under its promotion, the attack and defense of unmanned aerial vehicles will alternate in the future, and the means of mutual restraint will continue to innovate. (New Society)