The Embodied Best Paper team has stepped up in full force to tackle the perception-control integrated brain model.

Go-karting, a skill that not even all humans can master, has surprisingly been conquered by a bipedal humanoid robot??

Starting from the steps of "walking to the vehicle" and "sitting in the car", the robot perceives the road independently, controls the steering wheel, and presses the accelerator on its own.

It adjusts the applied force with its right foot, continuously corrects the direction with both hands, identifies track changes via visual perception, and constantly adjusts its center of gravity at the same time to maintain full-body balance.

Remarkably, this robot has fully mastered multi-contact balance, hand-eye-foot-brain full-body coordination, and extremely precise force control capabilities.

The company that developed this solution, Symbiotic Robotics, was founded just over a month ago.

Most strikingly, the bipedal humanoid robot in the video adopts an original end-to-end perception-control integration solution that leads the industry, which breaks through the long-standing separation contradiction of the mainstream "brain + cerebellum" architecture for current embodied intelligence. It no longer splits perception, decision-making and motion control into several relatively independent modules, but enables a single model to output the final robot action directly from the moment it perceives the surrounding world.

It is worth noting that end-to-end and Whole-Body Intelligence are the two most widely concerned technical paths in the global embodied intelligence academic circle in 2026. Few top laboratories and leading enterprises around the world have achieved relevant results, yet this cutting-edge technology has been successfully developed by a young Chinese company established for less than two months.

Who on earth is Symbiotic Robotics?

01. The "Dream Team" of Best Paper Winners in Embodied Intelligence

Symbiotic Robotics was founded at the end of June this year, less than 60 days from now.

If the first "unbelievable" point of this company is that it delivered an industry-leading perception-control integrated brain model just one month after its establishment, the second "unbelievable" point is its founding team.

The founding team of Symbiotic Robotics is almost a youth dream team composed of top Best Paper authors in the embodied intelligence field.

DING Pengxiang, CEO of the company, is a post-95s PhD from Zhejiang University and Westlake University, supervised by WANG Donglin. He has long been deeply engaged in embodied intelligence, with a series of dazzling achievements that are regarded as "legendary in the scientific research circle":

First author of the world's first quadruped VLA (QUAR-VLA, ECCV 2024) and the world's first humanoid VLA (Humanoid-VLA, Feb 2025), core author of ReconVLA (AAAI 2026 Best Paper), first author of the 50 million parameter lightweight benchmark VLA-Adapter (which gained more than 1k GitHub Stars in one week, now has over 2k Stars, ranking as the highest-starred open-source VLA project in China and the 5th VLA project with over 1k Stars worldwide), founder of the embodied intelligence open-source community OpenHelix Robotics (accumulatively over 5k Stars), and one of the top 5 most cited PhD researchers in embodied intelligence in China.

Other outstanding young talents in the core team are equally impressive.

All team members hold a doctoral degree, most are post-00s, and all have published papers at top academic conferences. It is a rare team (in China and even around the world) that has developed foundational models for quadruped robots, robotic arms and humanoid robots. Their achievements cover:

• First author of the CoRL paper that defeated Pi0.5 and won the first Best Paper at a top robotics conference for global Chinese teams;

• Author of the paper shortlisted for IJCNN Best Paper Candidate;

• First author of HEX, who proposed the first cross-embodiment Whole-Body Manipulation Model for full-size bipedal humanoid robots;

• First author of OpenWBC, one of the world's first open-source humanoid teleoperation systems;

• First author of Trajbooster, the first Loco-manipulation VLA that has attracted widespread industry attention, which greatly improves the data utilization efficiency across different embodiments...

Loco-manipulation, quadruped robots, robotic arms, humanoid robots, WBC data collection, pre-training, post-training, motion control, cross-embodiment...

Their solid capabilities and leading research are equivalent to gathering a group of authors of the most cutting-edge embodied intelligence doctoral papers in China over the past two years into one single company.

More importantly, their achievements are far more than just "publishing several papers". Three of the 5 most cutting-edge works in the international Loco-manipulation field are developed by this startup team with less than 20 people, whose Best Paper density is incredibly high.

A group of the youngest "first authors of top papers" in Chinese embodied intelligence have started to launch their own entrepreneurial business.

Interestingly, people who follow the embodied intelligence track closely will find that Symbiotic Robotics is not an isolated case.

In the past six months, a quiet "great migration" of talent paradigms has been taking place in the embodied intelligence track. Two years ago, the most typical entrepreneurial paradigm for Chinese embodied intelligence was that professors leave academic institutions to start businesses. University laboratories accumulate technology for several years, professors act as the technical banner of the company, and then enter the industry with their PhD, master students and laboratory achievements. Many star embodied intelligence startups follow this path.

However, starting from 2026, more and more young doctoral researchers who are at the forefront of scientific research have begun to start their own businesses.

Why? The only reason is that technology is developing extremely fast. In 2023, the industry was still discussing whether robots need "large models"; in 2024, VLA became the core concept; in 2025, dual-system models such as Figure Helix became popular rapidly; by 2026, the whole industry is focusing on the "end-to-end" direction.

This has led to an interesting "knowledge inversion" in the current industry: the most cutting-edge cognition of embodied intelligence technology may not be mastered by the "senior" industry veterans. The people who first know the next-generation technical paradigm may be the frontline doctoral students who process data, tune models, and test robots in person every day.

Which solution only performs well on Benchmark but fails when deployed on real machines? To find out where the most cutting-edge technology, talents and resources flow under the current technological trend, you need to ask the doctoral students who have just successfully run the latest models.

This change has already happened in the large model industry. LIANG Wenfeng, founder of DeepSeek, once introduced the profile of the core team in an interview with 36Kr: "They are all recent graduates from top universities, fourth-year or fifth-year doctoral interns who have not yet graduated, and some young people who have graduated for only a few years."

This situation exists not only in China, but also abroad.

It is true for large models, and it is also true for embodied intelligence.

02. End-to-end, the "Final Destination" of Embodied Intelligence

The most cutting-edge team with the most disruptive talent structure will naturally deliver the most cutting-edge and disruptive achievements.

As mentioned above, the subversion of Symbiotic Robotics lies in that it breaks the current mainstream "layered" solution, and directly targets the world-class difficult problem of end-to-end perception-control integration.

To understand its subversion, you need to know that in the past few years, the mainstream idea of humanoid robots is essentially "layered", or "brain + cerebellum". The upper-level brain is responsible for perception, understanding and planning, while the lower-level cerebellum is responsible for motion control and balance.

The team of Symbiotic Robotics is one of the first teams in China to propose this architecture, and it has published relevant results as early as 2023.

However, in the past two years, the problems of the "layered solution" have become increasingly prominent. Due to the natural indirect cascade error between the upper and lower layers, the mapping between perception and control cannot be directly connected, and the "brain" can hardly get the motion feedback from the "cerebellum", leading to poor task performance.

To make an analogy with human beings: when a person tries to catch a cup that is falling off the table, he will not first complete visual understanding, then pause to generate a trajectory, and finally move his arm. Instead, the moment his eyes see the cup falling, his body has leaned forward and his hand has stretched out. Perception, action and balance are inherently integrated.

For bipedal humanoid robots, this problem is even more extreme. Every time the robot raises its hand, turns around or bends down, its center of gravity will change; the change of foot contact force will in turn affect the movement of the upper body. The separation of upper and lower limbs, the inability to transmit task semantics to lower limb control and force control, long failure recovery links, and the difficulty in unifying data and goals...

This is why most of the current bipedal humanoid robot demonstrations on the market rarely show actions that require hand-eye-brain full-body coordination like Symbiotic Robotics' "go-kart driving" operation. It is not that they do not want to, but that they cannot achieve it.

The industry has recognized this problem. Since the second half of 2025, "Whole-Body Intelligence" and "end-to-end" have been recognized as the next development direction at top industry conferences, but few teams have delivered practical results so far.

Symbiotic Robotics is precisely targeting this direction.

The DPC (Direct Perception Control Model) developed by Symbiotic Robotics is a perception-control integrated foundational model for Loco-manipulation that breaks the barrier between the robot's "brain" and "cerebellum" and directly realizes end-to-end full-body coordinated operation.

Its input is multi-modal perception data (vision, language, Robot Stat, physical signals); through the Unified Perception Control Model, it directly generates full-body actions with force feedback from vision, language, ontology, historical feedback and contact feedback.

Finally, it can directly output full-body coordinated joint torque, targeting the "final destination" of the Interaction-Native Whole-Body Intelligence track.

It is particularly noteworthy that the Symbiotic Robotics team has taken the lead in verifying that multi-source heterogeneous data training can effectively realize full-body data scaling for humanoid robots, and its ontology-free pre-training algorithm can realize complex full-body coordinated operations with only a very small amount of real data.

Compared with traditional solutions, Symbiotic Robotics' DPC (Direct Perception Control Model) can optimize the perception and control signals of the brain and cerebellum synchronously (the hand-eye-brain full-body coordination when driving a go-kart), greatly improve the compliant force control ability of the robot (the pressing amplitude of the go-kart accelerator pedal and the vehicle speed), and realize the general adaptation of the same foundational model across different embodiments and configurations, with strong zero-shot generalization ability in real scenarios.

At present, "end-to-end" is the most cutting-edge consensus direction in embodied intelligence research, and it is also the evolution direction of the "next-generation robot brain" recognized by the industry in 2026.

In the past two years, the motion control WBC technology has become mature, the available data has exploded exponentially, and the route of the VAM perception and decision-making integrated foundational model has gradually converged. The "three major mountains" that once blocked the development of the embodied intelligence perception-control integrated brain have been gradually removed.

"In 2026, all three conditions are met. Now is the window period for the development of the bipedal humanoid end-to-end perception-control integrated brain, which is about 6-7 months, or even only 4-5 months. This window will not last long," DING Pengxiang, CEO of Symbiotic Robotics, told 36Kr.

Catching this window is very important; "winning once" is very important.

But that is not enough.

03. What Matters Is Not Only "Winning Once", But "Winning Continuously"

"All aspects of human society are designed for the human body structure, and many scenarios can only be realized by the bipedal humanoid configuration. From the perspective of first principles, bipedal humanoid robots must be the final form of the future," DING Pengxiang told 36Kr in the interview.

What Symbiotic Robotics is targeting is not short-term landing applications, nor any "low-hanging fruit", but a more ambitious and general future of embodied intelligence AGI.

This is undoubtedly a very cutting-edge, even very radical future.

But since 2022, the team of Symbiotic Robotics has almost bet correctly on all the technical development directions of embodied intelligence: from Humanoid-VLA providing task understanding, visual Grounding and action semantic entry, to TrajBooster providing data flywheel for end-to-end Whole-body Action Model, to HEX providing Humanoid-native State and Action representation basis for end-to-end model, and then gradually integrating the Force cerebellum / safety contact expert into the end-to-end model through distillation...

Winning once can be called "luck", but winning ten times in a row can never be explained by pure luck.

"Our team has been leading the development of bipedal humanoid Loco-manipulation. We are not only a participant in the industry, but also a maker of industry standards," said DING Pengxiang.

In 2026, Symbiotic Robotics is betting on the bipedal humanoid end-to-end perception-control integrated brain model, and believes that this will be the final destination of the industry.

Their "confidence" comes from their long-term in-depth cultivation in the embodied intelligence industry, never being limited by outdated methodologies, continuous breakthroughs and exploration, not only "winning once", but also "winning continuously".

From their experience, we can also see the quiet change of China's innovation paradigm.

For a long time in the past, China's technology industry was best at engineering and industrialization, making technology faster, cheaper and larger in scale. However, in the most cutting-edge foundational model and algorithm paradigm fields, the industry used to wait for the output of Silicon Valley first, and then discuss how China can catch up.

In the past few years, this rhythm has begun to change.

In the large model field, DeepSeek has proved that a group of Chinese researchers with a small average age can also give their own answers to the most cutting-edge problems in the world, and deliver breakthrough innovations from 0 to 1.

Now, a similar scene is taking place in the embodied intelligence field, and the protagonists are even younger.

Today, there are more and more "post-00s" players in the embodied intelligence track. They may be fourth-year or fifth-year doctoral students, or just over 25 years old, but they have already published the world's most cutting-edge papers, won Best Paper awards, defined the latest model architecture, and promoted the development trajectory of the entire industry.

What is really worth noting behind this phenomenon is not "how many new embodied intelligence startups have emerged in China", but that the gap between China's top young scientific research talents and the global innovation frontier is rapidly narrowing, catching up, even surpassing and leading the whole industry.

If the large model field made the world see the strength of young Chinese AI researchers for the first time in the past few years, the embodied intelligence field may be welcoming the next "world-class moment" belonging to China.

This time,