DaxAI Robot Horse Debuts at WRC'26: An All-Terrain Quadruped with 300kg Payload and 40km/h Top Speed

A "Robot Horse" Built for All Terrain

At the upcoming World Robot Conference (WRC'26), DaxAI is unveiling a remarkably ambitious product — an all-terrain robot horse. Unlike the bipedal humanoid robots or wheeled delivery bots we're familiar with, this machine employs a bio-inspired quadruped structure, carving out its niche in the "heavy cargo transport" domain.

Bio-inspired quadruped robots draw design inspiration from four-legged animals in nature — horses, dogs, cheetahs, and the like. Compared to wheeled or tracked mobile platforms, the core advantage of a quadruped structure lies in its adaptability to unstructured terrain. By selecting stable support points through discrete foot-ground contacts, it can traverse obstacles, climb stairs, and cross soft or uneven ground. The technological development of this field traces back to MIT's Marc Raibert lab in the 1980s, while the true engineering phase began with Boston Dynamics' BigDog project (2005). Current quadruped robot control typically revolves around two main approaches: Model Predictive Control (MPC), which optimizes gait in real-time based on physical models, and reinforcement learning, which develops robust locomotion policies through large-scale simulation training.

According to official specifications, this robot horse features the following core parameters:

• Range: Approximately 100 km / 10 hours of autonomous operation on a single charge
• Maximum Payload: 300 kg
• Top Speed: 40 km/h

These numbers represent an extremely aggressive positioning within the current quadruped robot landscape. This is no longer a lab-bound "electronic dog" bouncing around — it's explicitly targeting practical, heavy-payload, long-distance use cases like material transport and field operations.

The World Robot Conference (WRC) is a global robotics industry event hosted by the Chinese Institute of Electronics. Held annually in Beijing since 2015, it encompasses forums, exhibitions, and robotics competitions. WRC has become one of the largest robotics industry showcase platforms in China and the Asia-Pacific region, attracting hundreds of domestic and international companies each edition. For startups, WRC serves as a critical window for demonstrating technical capabilities, attracting investment, and reaching industry clients. DaxAI's decision to debut its product at WRC'26 reflects its market strategy of rapidly building brand recognition within China's robotics ecosystem.

The Technical Implications Behind DaxAI's Specifications

What 300kg Payload Really Means

Mainstream quadruped robots currently on the market, such as Boston Dynamics' Spot, typically carry payloads around 14 kg. Some industrial models from Unitree also fall within the range of a few dozen kilograms. Boston Dynamics' Spot is one of the most commercially successful quadruped robots today, priced at approximately $75,000, weighing about 32 kg, with a maximum payload of approximately 14 kg (using the official payload dock), around 90 minutes of battery life, and a top speed of about 1.6 m/s (approximately 5.8 km/h). Spot is positioned as a lightweight inspection and data collection platform, primarily serving power facility inspection, construction site monitoring, and hazardous environment exploration. Unitree's B2 sits higher on the spectrum, with a payload of approximately 40 kg and speeds up to 6 m/s.

DaxAI's claimed 300 kg payload is roughly equivalent to the carrying capacity of an actual working horse, placing extreme demands on structural strength, joint motor torque, and dynamic balance algorithms.

Quadruped robot joints are typically driven by high torque-density brushless DC motors paired with planetary gear reducers or quasi-direct drive configurations. Quasi-direct drive approaches use low gear-ratio reducers, preserving the motor's force transparency and back-drivability — beneficial for compliant force control but limited in torque output. For a 300 kg payload-class heavy quadruped, high gear-ratio configurations paired with force/torque sensors may be necessary to achieve precise ground reaction force control. On the dynamic balance algorithm front, Zero Moment Point (ZMP) theory and Virtual Model Control (VMC) are classic methods, while recent Whole-Body Control approaches based on full-body dynamics optimization can better coordinate multi-joint motion under heavy loads, preventing instability and toppling on uneven terrain.

Maintaining quadruped agility while bearing such weight typically requires more powerful drive systems and more robust mechanical frames. This explains why the product is named "robot horse" rather than "robot dog" — its positioning is fundamentally that of a transport tool, breaking entirely out of the existing "robot dog" category and entering territory traditionally covered by mules, small off-road vehicles, or military unmanned ground vehicles (UGVs).

The Real-World Challenge of 100km Range

Range has always been a pain point for mobile robots. Compared to wheeled locomotion, quadruped walking is inherently more energy-intensive, as every step involves leg lifting, ground contact, and balance adjustment.

From an energy consumption perspective, the high energy cost of quadruped walking stems from the fundamental characteristics of its locomotion. During each gait cycle, the legs must do positive work to raise the body's center of mass (against gravity), negative work to absorb landing impact (energy dissipation), and the swing legs consume energy accelerating and decelerating through the air. In contrast, wheeled locomotion on flat surfaces primarily overcomes rolling resistance, consuming an order of magnitude less energy. Strategies employed in the industry to improve quadruped energy efficiency include: Series Elastic Actuators (SEA) or parallel elastic elements to recover impact energy during ground contact; optimizing gait parameters (step frequency, stride length, duty factor) to minimize Cost of Transport; and using flight-phase gaits (such as trot and gallop) to reduce the number of simultaneously supporting legs, thereby lowering the load on each leg.

DaxAI's claimed 100 km / 10 hour figure, if accurate, implies significant optimization in energy management and locomotion efficiency — likely involving breakthrough designs in battery energy density (current lithium batteries range around 250-300 Wh/kg) and mechanical energy recovery.

However, it's important to note that range data is often measured under ideal conditions. Real-world scenarios involving full 300 kg loads, complex terrain, and frequent hill climbing would likely reduce actual range significantly. Such specifications need further verification through independent testing.

The Practical Boundaries of 40km/h Top Speed

A top speed of 40 km/h is also leading-edge among quadruped robots. This speed enables rapid repositioning across open terrain, suitable for emergency supply delivery and remote area resupply missions. However, high-speed locomotion presents an enormous challenge to the quadruped balance system. Whether it can reliably achieve this speed across real all-terrain environments is a key indicator of its technological maturity.

High-speed quadruped locomotion typically requires transitioning from walking gaits (walk) to trotting gaits (trot) or even galloping gaits (gallop), each corresponding to different leg coordination patterns and body dynamics characteristics. At 40 km/h, foot-ground contact time is extremely brief, compressing the time window available for the control system to perceive terrain and make adjustments. This places extreme demands on sensor response speed and control algorithm real-time performance.

Application Scenarios for All-Terrain Quadruped Transport Robots

The product logic of this robot horse creates an interesting differentiation from humanoid robots. While humanoid robots aim to "replace humans in human environments," quadruped transport robots aim to "access environments where humans cannot carry heavy loads."

Using quadruped robots for military logistics is not a new concept. DARPA's LS3 (Legged Squad Support System) project in the 2010s was a large quadruped transport robot developed by Boston Dynamics, designed with a payload of approximately 180 kg that could follow infantry squads through complex terrain to transport supplies. However, the project was ultimately shelved in 2015 due to excessive noise (from hydraulic drives) and battlefield practicality concerns. Since then, multiple Chinese companies and research institutions have also pursued similar directions. DaxAI's robot horse can be seen as a new attempt along this technological path in the era of electric drives — electric motor drives significantly reduce noise compared to hydraulic systems, though they still face challenges in power density.

Envisioned application scenarios include:

• Wilderness and Mountain Logistics: Rugged terrain impassable to vehicles in military logistics, geological exploration, and disaster relief
• Agriculture and Forestry: Transport tasks across complex terrain like terraced fields and hilly landscapes
• Emergency Rescue: Supply delivery and casualty evacuation assistance in areas where roads have been destroyed by earthquakes or mudslides
• Industrial Inspection: Long-distance patrol of large factory complexes and mining areas combined with sensor payloads

The common thread across these scenarios: complex terrain, heavy payloads required, and high range demands — precisely matching DaxAI's three core parameters.

A Clear-Eyed View of "Launch Event Specs": From Prototype to Mass Production

As a debut product at WRC'26, the information currently available still comes primarily from the manufacturer's marketing narrative. While it's natural to be excited by these impressive numbers, a measure of sobriety is warranted:

First, there is a chasm between specifications and mass production. Prototypes at trade shows and truly mass-deliverable, stably operating products often remain far apart. Many robotics products dazzle in demos but are slow to commercialize. The "concept-prototype-production" pathway in robotics is often far longer than anticipated. Take humanoid robots as an example: Tesla's Optimus went from concept unveiling in 2021 to still being in internal factory testing by 2024; Figure AI, despite massive funding rounds, has repeatedly postponed its mass production delivery timeline. The quadruped robot space is no different — Boston Dynamics took nearly a decade from Atlas research to Spot's commercialization.

Second, key metrics lack independent verification. Payload, range, and speed all require replication under standardized testing and third-party evaluation before the industry can truly trust them.

Third, cost and reliability are invisible barriers. Whether these performance levels can be achieved at reasonable cost, and the failure rate under prolonged high-intensity use, will ultimately determine whether this robot horse becomes practical or remains a concept demonstration. Core mass production challenges include: supply chain stability (particularly for high-performance motors, reducers, force sensors, and other critical components), quality consistency control, establishing after-sales maintenance systems, and reducing per-unit cost to ranges acceptable to target customers. For a 300 kg payload-class large quadruped robot, manufacturing precision of structural components, joint sealing protection ratings (particularly IP ratings for all-terrain use), and whole-machine reliability verification (MTBF metrics) are engineering challenges that must be conquered before mass production.

Conclusion: The Quadruped Transport Sector Deserves Attention

Regardless of the ultimate commercialization outcome, the emergence of DaxAI's robot horse reflects the robotics industry's trend from "showcasing technology" toward "delivering utility." While industry attention concentrates heavily on humanoid robots, the quadruped transport direction — focused on heavy payloads, long range, and all-terrain capability — may actually be the first to find solid commercial footing.

Its core value isn't about whether it looks like a horse, but whether it can truly replace human labor and traditional transport tools in those challenging "last mile" or even "last hundred meters" of difficult terrain. After WRC'26, how far this "robot horse" can run is well worth continued observation.