What’s the Point of a Humanoid Robot Jumping 2 Metres? ASIMO Explains Why Unitree’s “Superman” Matters
Unitree Robotics has recently showcased a high-performance humanoid robot currently referred to as “Superman”. In the company’s official test video, the robot is claimed to be capable of jumping roughly 2 metres from a standing position and reaching a top speed of 12.66m/s, or about 45.6km/h.
Numbers like these naturally invite comparisons with human performance.
But impressive running speed and jumping height do not necessarily mean humanoid robots are already close to replacing humans in everyday work.
Most current humanoid robots remain highly specialised machines. A robot that runs extremely fast does not automatically know how to move boxes, assemble components or respond intelligently when something unexpected happens on a factory floor.
The much harder goal is general-purpose capability: understanding the environment, interpreting a task and then using the same body to perform different kinds of work without being retrained for every individual situation.
So if humanoid robots are still far from becoming truly general-purpose workers, what is the point of building one that can run at about 45.6km/h or leap roughly 2 metres into the air?
One useful answer can be found in a much older robot: Honda ASIMO.
ASIMO was never really a “worker” either
For many people, ASIMO remains one of the most recognisable humanoid robots ever built.
Honda began researching bipedal robotics in the 1980s, eventually developing ASIMO. Over the years, the robot learned to walk, run, climb stairs, kick a football, carry objects and interact with people during public demonstrations.
Yet ASIMO never became the household assistant imagined in science fiction.
It did not enter millions of homes to cook meals, and it did not become a mass-produced factory worker. For much of its public life, ASIMO functioned more like a technological ambassador for Honda.
Viewed only as a commercial product, that might make decades of robotics research look difficult to justify.
But the technologies developed along the way did not disappear when ASIMO stopped making regular public appearances.
Research into control, perception, visual recognition and autonomous movement gradually found applications elsewhere across Honda’s mobility technologies.
ASIMO learned not to fall. Cars learned to control their motion
One of the most basic and difficult challenges in bipedal robotics is balance.
Humans walk without consciously calculating every movement, but a robot must constantly estimate its posture, centre of gravity, foot contact and the amount of force required at each joint.
It must understand where its body is moving and correct itself almost instantly.
The same fundamental idea can be applied to vehicle motion control.
Honda has said that the Honda 0 SUV will use high-precision attitude estimation and stability control based on 3D gyroscope technology, drawing on knowledge accumulated through the company’s robotics research.
ASIMO had two legs to control. A car has four wheels.
But both systems need to understand how a moving body behaves in multiple directions. A vehicle must monitor pitch, roll and yaw, then coordinate different systems according to tyre grip and driver inputs.
In other words, one of the central engineering questions behind ASIMO was simple to describe but extremely difficult to solve: how do you keep a moving machine stable?
More than two decades later, that same question still matters in automotive engineering.
ASIMO’s “eyes and brain” did not disappear either
ASIMO research eventually went far beyond simply walking and running.
Honda also had to teach the robot to recognise its surroundings, detect people and movements, and make decisions based on what was happening around it.
Those challenges have obvious parallels with modern ADAS, autonomous driving, sensor fusion and environmental perception.
Honda has even revived the ASIMO name for the software platform used by its next-generation Honda 0 Series vehicles: ASIMO OS.
This does not mean Honda simply transferred software from the old ASIMO robot directly into a modern car.
Instead, the name reflects how ideas developed through years of robotics research — including control, perception and autonomous decision-making — continue to influence Honda’s next generation of software-defined vehicles.
Seen from that perspective, the importance of Unitree’s “Superman” becomes less about the headline number and more about the engineering required to achieve it.
Today it makes a robot jump. Tomorrow the technology may go somewhere else
Making a humanoid robot leap roughly 2 metres from the ground and land in a controlled manner requires far more than simply fitting a stronger motor.
The system must coordinate high-power-density motors, joint output, battery discharge, thermal management, posture sensing, real-time control and the enormous impact forces created during landing.
All of those systems have to work together within fractions of a second.
Today, that engineering may be used to make a humanoid robot jump. But the most commercially valuable result in the future may not be the “Superman” robot itself.
High-power-density motors can be used in other mobility products. Better posture estimation can contribute to vehicle control. Joint-control technologies may find applications in industrial equipment or prosthetics, while advances in battery and thermal management can move into other high-performance electromechanical products.
ASIMO already demonstrated that research developed for robots does not necessarily stay inside robots.
What humanoid robots still lack is general-purpose intelligence
Humanoid robots are progressing quickly.
Some can run faster, some can carry boxes, others perform flips, and several are already being tested inside factories.
But many of these demonstrations share the same limitation: the robot has usually been trained extensively for a specific task.
A human worker is different.
A person can move a box, notice an unexpected obstacle and walk around it. If a tool falls to the floor, they can pick it up. If the workflow changes, they can understand a new instruction and adapt.
Combining perception, common sense, physical control and real-time judgement into one flexible system may be the much bigger challenge humanoid robotics still needs to solve.
So when a robot jumps around 2 metres into the air, perhaps the most useful question is not: “Whose job can it replace tomorrow?”
A better question may be: “What engineering problems had to be solved to make that jump possible?”
More than 20 years ago, people watched ASIMO run, kick footballs and wave at crowds without knowing where all that research might eventually lead.
Today, Honda has carried the ASIMO name into the software and control technologies of its next-generation vehicles.
Unitree’s “Superman” may leave behind something similar. Its most important legacy may eventually be much more than a 2-metre jump.
Humanoid robots are still a long way from becoming truly general-purpose machines that can work like humans. But that does not mean today’s running, jumping and acrobatics are merely stunts.
ASIMO showed how research with no obvious short-term commercial application can eventually contribute to control systems, perception and software used in cars and other mobility products.
The most interesting thing about a robot jumping 2 metres is therefore not just the height. It is the collection of engineering problems that had to be solved to make that single movement possible.
Comments
Post a Comment