Don't let the robot dogs go, but launch an attack under the shadow of the giants.

The gap before entering the "free - for - all stage" of robotic dogs is that Unitree Technology firmly holds 70% - 80% of the market share. Without even discussing how competitive Unitree Technology is in the robotic dog field, one can get a sense just by looking at the fact that the current robotic dogs on the market look very similar to Unitree's robotic dogs. To some extent, Unitree Technology has become strongly associated with the robotic dog race.

The story of the "duel between two robotic dog giants" of Unitree and Deep Robotics is becoming a thing of the past.

On the side of humanoid robot manufacturers, companies like Zhipu AI and Zhongqing Robotics have joined the fray; start - ups in the robotic dog field such as Jüwei Technology and Zhishen Technology are emerging; in the consumer market, companies like Weilan Technology and Hengzhi Future are rising rapidly; Yuejiang, the "first stock in collaborative robots", has also entered the battlefield. The influx of diverse participants may drive the long - stable robotic dog market into a "reshuffle period".

The reasons for entering the robotic dog market are not hard to understand. Whether it's the technological investment, short - term commercial returns, or long - term value, it's an economic equation that can be clearly calculated.

Firstly, the technology of robotic dogs is relatively mature. After the technology of quadruped robotic dogs was open - sourced, both predecessors and new entrants can start from a relatively high technological starting point. While lowering the entry threshold, they can also launch a relatively competitive product into the market and secure commercial orders.

Secondly, robotic dogs have become "standard products" with huge potential for secondary development. Whether it's adding a robotic arm to enhance operation ability, or adding wheels or extra legs to improve off - road capabilities and expand application scope, it all indicates a vast commercial future. Even if they can't serve as productive forces in real - world workplaces, they can still secure procurement orders in the scientific research and laboratories of high - efficiency enterprises and large companies.

Finally, from a commercial perspective, the application scenarios of robotic dogs are relatively homogeneous, mostly for industrial inspection, security, and logistics inspection, mainly focusing on mobility. Although this is a good place for robotic dogs to be implemented at present, it actually reveals the limitations of early - stage technology. When the technology makes a leap, the commercial scenarios will naturally be more diverse, with more commercial opportunities to be explored and deep - dived into, and the market growth will be sufficient to support several enterprises.

It can be seen that investing in robotic dogs is a relatively reasonable ROI investment, so naturally no one wants to miss out on the robotic dog market.

However, the gap before all players can enter the "free - for - all stage" of robotic dogs is that Unitree Technology firmly holds 70% - 80% of the market share. Without even discussing how competitive Unitree Technology is in the robotic dog field, one can get a sense just by looking at the fact that the current robotic dogs on the market look very similar to Unitree's robotic dogs. To some extent, Unitree Technology has become strongly associated with the robotic dog race.

Snatching food from the jaws of a powerful tiger may be the current proposition of this market. Judging from the actions of industry players, an obvious solution is not to go head - to - head with the giants, but to attack under their shadows. It could be finding niche scenarios or focusing on differentiated technologies. In any case, they can secure an ecological niche.

Quadruped Robotic Dogs Are Truly Appealing

Developing quadruped robots has become a rare consensus among embodied intelligence enterprises. It can be divided into two categories: one is enterprises that "turn back" to develop quadruped robots after working on humanoid robots, and the other is start - ups in the embodied intelligence field that plan to start with quadruped robots first. Behind this are two logics: the former is a short - term detour in the long - cycle development of humanoid robots, while the latter sees great potential in the quadruped field.

For the former, who actively "down - grade" after working on humanoid robots, developing quadruped robots is the most immediate strategic move.

Image source: Zhipu Robotics

Firstly, the market education cost for quadruped robots is the lowest. Quadruped robots are a transitional form that customers "can understand and are willing to order". Whether it's due to the 10 - year PR campaign of Boston Dynamics for "robotic dogs", or customers regarding quadruped robotic dogs, especially wheeled ones, as "self - driving cars", the cognitive threshold is extremely low.

Moreover, large scientific research and government customers have strong stickiness. Especially for those humanoid robot enterprises invested by cooperative universities and local governments, if their technology portfolio only includes humanoid robots, they will miss out on the repurchase market of these customers.

Secondly, the stories that capital values have changed. They are no longer willing to pay for simple humanoid demos and require to see "self - hematopoietic ability". Therefore, humanoid robot enterprises must first tell a "profit - making story" before talking about the "trillion - dollar future". The high gross margin and large number of real orders of quadruped robots can just fill the "bottomless pit" of the money - burning humanoid robot business. It is understood that the gross margin of quadruped robots is generally high, and some enterprises can even maintain it at over 60%.

Compared with humanoid robots, which usually have orders of only dozens or hundreds of units, quadruped robots are the only form that can achieve large - scale shipments and generate positive cash flow at present. Unitree is a good example. Whether it's the shipment of 23,700 quadruped robots in 2024 or its ability to remain profitable since 2020, Unitree has set a model, proving that the commercialization path of quadruped robots is feasible.

For start - ups, developing quadruped robotic dogs not only allows them to catch the wave of embodied intelligence but also enables them to enter the "sweet spot" with low entry costs, implementable scenarios, and easily - accepted products.

Image source: Jüwei Technology

For example, in 2018, MIT open - sourced the Cheetah Mini, including the "full control stack + motor driver board", allowing new entrants to skip the stage of "how to prevent the dog from falling". Full - set SLAM + whole - body MPC such as ROS2 Navigation2, OCS2, and LIO - SAM can be directly used.

In an interview with GeekPark in 2021, Wang Xingxing said, "The foreign open - source has made it easier to develop robotic dogs from scratch. A robotic dog that used to score 60 points may now reach 80 points."

As the overall technology matures, the proposition for new players has directly shifted to "productization", including identifying user needs and improving product industrial design. This is also why many founders of quadruped robot start - ups have a background in product management, such as Zhao Zhelun, the co - founder of Vita Power, and Liu Yulong of Zhishen Technology.

This also means that a mature technological background not only reduces R & D costs but also speeds up the pace of entrepreneurship. It allows for the preparation of scenario verification in a relatively short time, directly pointing to the future commercialization direction.

In terms of scenario implementation, quadruped robots have penetrated into many scenarios and can undertake a large number of real - demand orders, thanks to their all - terrain adaptability, modular payload advantages, and industrial - grade reliability.

For example, in the B - to - B market, Unitree Technology and Deep Robotics have long been involved in power grid inspection; Lingtian Intelligence, the "first emergency response stock", provides emergency response robots for all types of disasters such as fires and floods; Shenhao Technology has developed security dogs for the police system. In the consumer market, scenarios such as "guide dogs", "companion robots", and "smart housekeepers" are also real and represent a broader commercial space.

Overall, the development of AI has also brought revolutionary opportunities to robotic dogs. With the support of technologies such as pure - vision solutions and operation algorithms, robotic dogs can not only provide mobility and load - bearing functions but also develop operation capabilities.

With low costs, wide application space, and diverse forms, the alluring robotic dogs have naturally become a "scarce product" in the eyes of all sectors.

Our Development of Robotic Dogs Is Less Than "1%"

Looking back, it's difficult to find a clear time point, but we can clearly sense that the development of robotic dogs "paused" for a few years.

This "pause" is multi - dimensional. There has been no major technological breakthrough, and there have been no new stories in commercial implementation. This is not the problem of the participating players but a short - term stable period under the backdrop of overall technological development. After all, technology doesn't advance overnight, and changes in production relations also require time to settle.

Fortunately, with the influx of new players and major technological leaps, robotic dogs have finally made new progress.

For example, Jüwei Technology has found application scenarios in outdoor mountain climbing and camping. The fully autonomous following quadruped robot MOVENEWT1 developed by the company can achieve 1.5 times the power output in the same volume and has a battery life of up to 12 hours, enabling long - term and high - intensity outdoor sports.

At the same time, it has local intelligent decision - making ability. With a built - in high - computing - power processing platform, it can complete environmental data analysis, task planning, and human - machine interaction responses without relying on an external server. In mountain - climbing scenarios without network coverage, it can generate navigation paths in real - time and activate the photography module. Moreover, it can be equipped with a refrigerator module to become a "mobile cooler" for camping. It is understood that the company has received cooperation intentions from several outdoor brands in Hangzhou.

Image source: Jüwei Technology

In addition to outdoor scenarios, companies like Zhishen Technology are developing intelligent guide dog products, and Shenhao Technology is promoting applications in the field of smart agriculture, such as using robots to transport tea on tea plantations.

Beyond the in - depth exploration of scenarios, a broader industry impact is that the industry is now engaging in real - scale competition. In the past, it was just minor skirmishes, but now large - scale shipments will begin, ultimately driving down costs across the entire supply chain.

It's worth mentioning that the entry of cross - industry players has also brought "new things" to the robotic dog market. For example, Yuejiang has not only broken the traditional quadruped form and presented a more stable hexapod form but also constructed a four - dimensional synergy of hardware - algorithm - scenario - ecosystem based on its own advantages, bringing new solutions to the market.

Looking to the future, two directions of evolution are becoming increasingly clear from the current actions.

Firstly, the types will become more and more segmented. At present, most robotic dogs have a wide range of applications. "Being all - encompassing" is the shortest path to accumulate resources in the early stage, but it's definitely not the final form of the market structure. As more players enter the market, the gradual segmentation of applications is an inevitable trend, as each company has different backgrounds and injects different new impetuses.

Humanoid robots have provided a typical example. After companies like YouiTech and Qinglang launched humanoid robots, more segmented humanoid robots have naturally emerged based on their respective backgrounds. For example, Qinglang has developed humanoid robots that are more suitable for hotel service scenarios based on the real - world data accumulated over the years in the hotel industry.

Robotic dogs will also follow the evolutionary path of humanoid robots and move towards more vertical applications based on the backgrounds of new players. For example, Kangdi Technology, which has established a foothold in the overseas market with off - road vehicles, is now collaborating with Deep Robotics to develop caddy robotic dogs for golf courses. On the one hand, it provides a new application for robotic dogs; on the other hand, it can demonstrate product strength through the penetration of vertical scenarios.

Secondly, on the basis of segmentation, robotic dogs will also play the role of the "central nervous system". This has to do with the nature of embodied intelligence. The mission of embodied intelligence products is to be the carrier for AI to interact with the physical world. Under this premise, embodied intelligence products such as robotic dogs can interact with other AI software without hardware carriers, releasing ecological value.

An example can illustrate this more clearly. In the smart home scenario, in the current mature smart home solutions, smart speakers often serve as the "central hub" of home AI. Users issue commands through voice interaction, and the smart speaker controls the opening and closing of curtains, lighting, etc.

Now, robotic dogs can take over the role of smart speakers. After receiving human instructions, they can control home lighting, curtains, etc. They can also take care of children, the elderly, and pets at home through visual perception and autonomous navigation capabilities and report emergencies in a timely manner. This is not just a fantasy. Xiaomi developed a product several years ago. Its robotic dog, Tie Dan, has Xiaomi's voice assistant built - in. Users can control smart home devices at home, such as turning on the TV and closing the curtains, by saying "Tie Dan, Tie Dan".

It can be seen that the future commercial imagination space for robotic dogs is vast, and this is the area where Unitree Technology and Deep Robotics have not established a decisive leading advantage, providing opportunities for new entrants.

This article is from the WeChat public account "Embodied Intelligence Research Society". The authors are Peng Kunfang and Lü Xinyi. It is published by 36Kr with permission.