For some time now, many countries around the world have been focusing on developing unmanned combat capabilities and accelerating the deployment of combat readiness. In this process, different countries have formed two path choices for the construction of unmanned combat forces: "embedded integration" and "independent military formation" - the former pursues short-term breakthroughs, while the latter focuses on long-term system integration. The coordinated linkage of the two paths is a necessary stage in the construction of unmanned combat forces. With technological breakthroughs and strategic demand adjustments, the scope of application of the "independent military" model is expected to gradually expand in the future, while the "embedded fusion" model, as a safe choice during the technological transition period, will still be the dominant path for most countries. Differences in demand lead to path differentiation. Currently, countries have reached a basic consensus on important issues such as distributed deployment of unmanned combat forces, autonomous operations, and manned/unmanned collaboration. However, based on their respective strategic needs, technological foundations, and other conditions, there are significant differences in development concepts. The top-level planning and design present demand differentiation. In its "Joint Global Command and Control" architecture, the United States explicitly regards unmanned systems as key nodes linking various domains, emphasizing the realization of global perception, multi domain strikes, and cross domain collaboration in unmanned combat in system confrontation. Russia, on the other hand, regards "independent military formation" as a short-term priority and plans to complete the formation of unmanned system units within the year, hoping to solve the problems of dispersed forces and inefficient coordination through independent organization. Japan plans to develop the application of multi aircraft collaborative command tactics under intelligent empowerment; Israel established the Artificial Intelligence and Autonomous Systems Authority at the beginning of this year to coordinate and promote the integration and application of intelligent technology and unmanned systems. Selecting the impact path of institutional innovation. The path of "forming an independent army" is centered around highly professional centralized command. For example, Ukraine set up an unmanned system force in May last year, with existing UAV companies/battalions as the backbone, supplemented by UAV units at the brigade company level, to realize the centralized deployment of unmanned forces. After establishing a drone force at the beginning of this year, Poland has been promoting the research and development of domestically produced unmanned systems while procuring foreign equipment, accumulating technical experience for the generation of military capabilities. The 'embedded fusion' path emphasizes the deep integration of unmanned systems with traditional military forces. The "Titan" program proposed by France aims to create an expeditionary force with human-machine collaboration and fully embedded unmanned systems in traditional ground force structures. In addition, modular task grouping serves as an intermediate form between the two paths, with task driven flexible grouping as the core for experimental application. For example, NATO temporarily formed an experimental drone/boat hybrid team during exercises to implement coordinated operations with aircraft carrier strike groups. With the increasing strategic weight of unmanned combat forces in modern warfare, their role has surpassed auxiliary support and become a key factor in determining the outcome of the battlefield. Multiple countries have achieved dynamic adjustments in equipment, organization, and tactics through practical exercises and training, enhancing the effectiveness of unmanned combat systems. Practice and validation drive the iteration of technical routes. Performance training is an important part of verifying technical routes and exposing system shortcomings. In this year's US military "Convergence Project - Keystone 5" exercise, unmanned cluster collaborative operations exposed problems such as "insufficient response speed and data delay of command and control networks". The US military immediately introduced artificial intelligence to promote solutions. In the live fire coordination of unmanned aerial vehicles during the "Fuji Comprehensive Fire Exercise 2024" by the Japanese Self Defense Forces, issues such as "incompatible data link protocols" prompted the launch of the "Unified Communication Standards and Interoperability Enhancement" project. From discovering shortcomings to promoting upgrades, joint exercises have steadily evolved unmanned combat from "theoretically feasible" to "practically usable". Practical testing has forced adjustments to organizational tactics. On the battlefield between Russia and Ukraine, the Ukrainian military's unmanned aerial vehicles initially encountered repeated strikes due to the lack of data link coordination. Starting in 2024, they adopted an embedded mode of "unmanned aerial vehicle company/battalion traditional brigade" to shorten the kill chain and reduce redundant strikes, which confirmed the tactical value of the embedded mode; The Russian military solved the problem of air defense suppression with the "unmanned platform cascade configuration", first using patrol missiles to consume enemy combat power, and then using unmanned tanks to infiltrate. This tactic was quickly promoted after practical verification. The shortcomings exposed by unmanned combat methods in conflicts have forced the optimization of organizational structure; The adjusted battlefield effectiveness, in turn, verifies the adaptability of the new tactics in reverse, forming an upgraded loop of "problem discovery plan adjustment actual combat verification". The practical difficulties need to be adjusted and solved. Similar to other military transformations, the construction path of unmanned combat forces requires finding the optimal solution in the dynamic balance of multiple variables. Both the "independent military formation" and "embedded integration" models are currently facing many bottlenecks. At the technical level, the "independent military" mode requires the high autonomy of unmanned systems as a prerequisite, requiring equipment to have complex environment perception, multi-target dynamic allocation, cross domain collaborative communication, and other capabilities. If the technology maturity is insufficient, there may be a significant gap between actual performance and expectations. The Russian unmanned combat vehicles have exposed problems such as insufficient communication anti-interference capability and remote control delay in actual combat, which seriously affects execution efficiency. In this case, the "embedded fusion" mode is still a better choice for reducing decision risk. In terms of cost, although the "embedded integration" model can rely on existing systems to reduce construction costs, the inertia of traditional military units may pose obstacles to integration. Some unmanned projects of the US military have experienced delays in milestones due to issues with reliability review and tactical doctrine adaptation; Although Türkiye has translated the actual combat experience of TB-2 UAV into preparation specifications, equipment delivery was delayed due to supply chain fluctuations. The "independent military" model requires large-scale deployment of unmanned systems and high investment requirements. Some countries have to reduce their scale due to economic pressure and turn to building "small and sophisticated" unmanned combat independent units. In terms of talent, manned/unmanned collaborative combat requires officers and soldiers to be able to operate unmanned combat platforms while also completing tasks such as battlefield situation assessment and combat mission planning. In early September, South Korea established the "Light Drone and Anti Drone Task Force" and announced a large-scale training program for drone operators. Currently, in the "embedded integration" mode, the training of unmanned combat personnel in some countries is more focused on equipment operation; The talents required for the "independent formation of the army" model need to undergo systematic training and practical exercises to form true combat capabilities. From the current situation, there are still shortcomings in the training of unmanned combat personnel in various countries. Overall, the "embedded fusion" model relies on traditional systems to reduce technological risks, while the "independent formation" model unleashes the potential of scale through innovative organization. The two models are not substitutes, but complementary forms adapted to different levels and scenarios. In the future, the construction and development of unmanned combat forces in various countries may present a dual track trend, constantly maturing in dynamic balance. (New Society)