This year marks the 10th anniversary of the World UAV Conference and Shenzhen International Low-Altitude Economy Expo. At the exhibition site, if you're still focused on counting the number of drones on display or comparing which one flies higher or faster, your perspective is already outdated.

As a seasoned veteran in the low-altitude economy sector, I've observed that the most significant shift at this year's expo is the evolving 'soul' of low-altitude equipment. Embodied intelligence is propelling the transformation of low-altitude equipment from mere 'aerial vehicles executing preset instructions' to 'intelligent agents endowed with spatial awareness and autonomous decision-making capabilities.'

When 'flying in the sky' and 'operating on the ground' start to share a common cognitive framework, the large-scale commercialization of the low-altitude economy finally finds its true value proposition, yet it also encounters the most formidable application challenges.

Cognitive Synergy Between Air and Ground

Historically, drones operated like 'blind men feeling an elephant,' relying solely on flight control systems to navigate environmental changes. Now, embodied intelligence equips these devices with 'spatial common sense' and autonomous decision-making prowess.

1. Aerial Equipment: From 'Obstacle Avoidance' to 'Semantic-Level Interaction'

At the expo, industry leaders such as DJI Innovation, Autel Robotics, JOUAV, Chengzhi Intelligence, Damoda, Ehang, TopXGun Robotics, Meituan UAV, and Yunsheng Intelligence have shifted away from competing solely on obstacle avoidance sensors and price advantages. Instead, they showcased comprehensive 'air-space-ground integration' solutions powered by cutting-edge technologies.

Autel Robotics aims to create an 'Air-Ground Integrated Cluster Intelligence' system, utilizing Physical AI Agents (such as the new-generation full-stack AI flagship drone Autel EVO III series, the versatile new industrial drone Autel Apex, and the Autel Dragonfish-25 AI full-stack equipment matrix) as execution terminals, and Digital AI Agents—the EVO Claw cluster command and control system—as the decision-making core.

This architecture seamlessly integrates situation awareness, decision recommendation, mission planning, simulation, cluster collaboration, execution, and reflective evaluation, enabling multi-device cluster linkage (coordinated operation), comprehensive situation awareness, and full-process automation, transitioning from standalone operations to systematic intelligent decision-making.

Currently, this architecture precisely targets three high-value scenarios: smart security, transportation, and energy, addressing traditional inspection challenges such as high risk, low efficiency, and data silos, transforming intelligent clusters into truly schedulable, collaborative, and reusable productivity tools.

Damoda focuses on low-altitude cluster embodied intelligence technology, demonstrating how it reconstructs drone performances through hardware platforms, automated deployment, intelligent scheduling, content generation, and on-site operations for drone cluster shows.

At the hardware level, the intelligent vehicle-mounted system 2.0 enables industrialized operations with 'stop-and-fly' capabilities, while the X1 special effects drone expands aerial visual boundaries. At the cognitive level, the AI script editor replaces cumbersome manual processes, compressing creative cycles from 'weeks' to 'minutes,' solving the 'what to do' decision-making problem. At the execution level, precise AI agent implementation forms a 'perception-decision-execution' closed loop. This full-stack capability, covering 'hardware-scheduling-content,' essentially transforms drones from 'passive execution' to 'active service,' providing critical support for the low-altitude economy's evolution from project-based to industrialized operations.

Emerging companies such as Grid Intelligence Computing, Embodied Intelligent Navigation, and Amov Labs leverage their strong technical advantages to fill gaps in complex extreme operating scenarios with weak or no signals.

Grid Intelligence Computing: As a physical AI enterprise deeply rooted in the industrial sector, it centers on original generalized vector space theory, offering core advantages of micro-data, low computing power, high precision, and strong generalization, potentially breaking through AI vision solution bottlenecks. Its self-developed modules address application pain points of high precision, large computing power, high power consumption, and high latency caused by data fusion from multiple devices in unstructured environments, significantly lowering application thresholds and costs. Currently, drones equipped with this module have been applied in forestry inspections, initially completing a 'theory-technology-single scenario' closed loop.

Embodied Intelligent Navigation: As a core technology representative in satellite-denied and passive navigation and positioning, it independently developed an edge-side physical AI model, integrating artificial intelligence with traditional avionics systems to reduce reliance on human operators in daily operations. Its pure vision (passive) real-time perception technology breaks through external perception pain points in 'confined spaces,' adaptable to various equipment configurations such as drones, unmanned vehicles, and robots.

Bosheng Robotics: As a pioneer in the flying robot industry, the company showcased its latest full-stack self-developed FlyCore fusion control system, addressing industry pain points of flight stability in denied environments. FlyCore uses 'vision + laser + IMU + RTK' multi-source fusion technology for positioning, combined with a 'left-right brain + cerebellum' computing power physical isolation architecture, ensuring sufficient computing power for SLAM mapping and path planning under high loads, enabling all-weather stable autonomous flight without GNSS.

Embodied intelligence endows aerial vehicles with 'continuous environmental awareness' capabilities, enabling autonomous planning of optimal interaction paths and autonomous operation in denied or extreme environments.

2. Ground Equipment: The 'Cerebellum' Evolution for Air-Space-Ground Collaboration

From the essence of the low-altitude economy concept, it is never an isolated aerial endeavor but rather a low-altitude activity emphasizing systematic operations and value generation. In the exhibition hall, the evolution of robotic dogs is equally rapid, becoming core autonomous carriers for air-ground integrated collaboration.

Xuanji Power debuted its heavy-duty quadruped robot Hypertron T01, showcasing its high torque density, strong load capacity, all-terrain obstacle crossing, and unmanned operation capabilities.

Designed for heavy-duty tasks, the Hypertron-T01 quadruped robot features an IP67 protection rating, operates in temperatures ranging from -20°C to 55°C, and can continuously climb 25 floors with a 60kg load. It has passed high-temperature and low-temperature extremes, anti-vibration transport tests, and can operate continuously for 4 hours under full load.

With all-terrain mobility, heavy-load operations, and all-weather unmanned operation advantages, it supplements the blind spots of 'aerial eyes' and drone perspectives, quickly penetrating core scenarios such as power inspections, emergency rescue, and public security. It has already been deployed in bulk and market-validated by leading clients such as China Southern Power Grid.

Strawberry Innovation: As one of the few alternative companies adhering to long-termism in the industry, its self-developed SuperDock automatic airport (charging version) ranked second only to DJI Dock 3 in the 72-hour airport endurance race hosted by State Grid Intelligence. At this expo, Strawberry Innovation showcased its full matrix of products and solutions, including the SuperDock M400 airport (charging/battery swap version), M400 airport (vehicle-mounted version), M4 battery swap version, and M400 (logistics version).

Without reliable ground-based embodied intelligent terminals, 'air-space-ground integration' remains a half-baked solution. Embodied intelligence equips ground equipment with autonomous navigation and operation capabilities in unstructured terrains, enabling data sharing and task complementarity with drone equipment.

From a scenario perspective, drones perform wide-area scans in the air to lock targets, quadruped robots autonomously plan routes to quickly reach and verify targets, while cloud-based embodied intelligence hubs coordinate uniformly, forming an 'aerial reconnaissance-ground execution-cloud judgment' closed loop. This enables the low-altitude economy's three-dimensional collaboration to transition from demonstrations to industrial practical applications.

Commercialization: No Longer 'Finding Nails with a Hammer'

In previous years, the low-altitude economy faced criticism for 'taking off but not landing.' From a fundamental technology perspective, the root cause was that machines were not intelligent enough, leading to low efficiency and high manual operation costs.

Embodied intelligence represents the most significant variable in the low-altitude economy. No longer reliant on pre-set flight paths by human operators, they can autonomously plan routes, identify targets, and even execute 'air-ground integrated' collaborative operations in complex environments. This signifies the upgrade of aerial vehicles from 'data-collecting eyes' to 'direct-working helpers,' reshaping the commercial logic of large-scale operations:

Anchor Point 1: Marginal Costs Approach Zero, Breaking Through Long-Tail Scenarios

Scenarios like power inspections and emergency rescue have struggled to scale due to numerous long-tail problems (such as sudden obstructions, fog, heavy rain, and other extreme weather) in these high-frequency application scenarios, making them reliant on skilled human operators. Embodied intelligence equips equipment with generalization capabilities, allowing one model to handle thousands of unexpected situations, potentially eliminating human operator costs and finally crossing the scalability threshold for low-altitude operations' ROI (return on investment).

Anchor Point 2: Air-Space-Ground Integrated Closed Loop, Efficiently Completing the 'See-to-Do' Transition

Previously, aerial vehicles were only responsible for 'seeing,' requiring personnel to carry various equipment to 'do' after identifying problems. Embodied intelligence-driven air-ground collaboration truly achieves efficient 'detection and disposal.' For example, in forest fire prevention, drones lock fire points, and firefighters carry fire-extinguishing robotic dogs (or robotic dogs carrying fire-extinguishing equipment) to suppress fires, significantly reducing deployment costs and personnel casualties.

The formation of a service closed loop means the low-altitude economy no longer sells 'flight services' but 'results,' with customer unit prices and willingness to pay expected to rise exponentially.

The Harsh Reality of 'Usable' to 'Practical'

While future trends are exciting, commercialization still faces harsh realities. Behind the glossy demos at the expo lie three major pain points for embodied intelligence in the low-altitude domain:

Pain Point 1: The 'Deadlock' Between Computing Power and Payload

Embodied intelligence requires significant computing power, but low-altitude equipment is highly sensitive to weight and power consumption. Stuffing a box with hundreds of TOPS of computing power into a drone directly halves its endurance; using edge computing solutions struggles to support end-to-end large model inference.

Additionally, payload and computing power remain a seesaw relationship. Until chip fabrication processes and model lightweighting breakthroughs occur, commercialization will remain limited.

Pain Point 2: The 'Gap' Between Simulation and Reality

Low-altitude environments are highly dynamic and unstructured. Embodied intelligence heavily relies on data feeding, but extreme long-tail data in low-altitude environments is extremely scarce, making the transition from simulation to reality highly costly.

Drones that perfectly avoid obstacles in simulators at the expo may appear 'intellectually disabled' when faced with dense anti-theft nets in urban villages and erratic flying kites.

Pain Point 3: The 'Mute' Dilemma of Infrastructure

Decision-making by embodied intelligent agents requires training on environmental information. However, current low-altitude infrastructure mostly solves 'communication' and 'surveillance' issues, lacking 'semantic-level environmental modeling' for machine intelligence. If cities cannot provide intelligent equipment with a readable 'digital map traffic light' system, low-altitude intelligent agents will remain isolated and blind, making scalability merely theoretical.

The signal from the 2026 World UAV Conference is crystal clear: the second half of the low-altitude economy will not be won by who flies higher but by who thinks faster and operates more stably.

Embodied intelligence is the only antidote for low-altitude equipment to cross the chasm from 'usable' to 'practical,' reshaping the commercial foundation of the low-altitude economy. However, we must also soberly recognize that from 'intelligent agents' in labs to 'productive forces' shuttling in the skies, three mountains—computing power, data, and infrastructure—still stand in the way.

The wind is blowing in the right direction, but the ship must still be built. Talk less about disruption and focus more on solving real-world pain points, and the trillion-dollar blue ocean of the low-altitude economy will truly open up to embodied intelligence.

*Editor's Note: Original content is hard-earned; please respect the author. For reprints, please contact us.