logo metameha

Unitree G1 humanoid robot can now run (but autonomy lags)

2 Feb 2025 | Robots, Artificial Intelligence | 0 comments

Videos of the Unitree G1 humanoid robot show clear gains in how it moves. It can run outdoors and walk with a smoother gait. The clips look exciting, but they do not prove full autonomy in messy, unknown places. For buyers and fans, the key is to separate agility from navigation and task skills.

G1 humanoid robot running: what the new clips show

Recent event videos from 2025 capture G1 running on campuses and show floors, with fewer arm flails and steadier footfalls than earlier demos. These are event demonstrations, where teams can select routes and manage risks. See examples: USC/RSS 2025 outdoor run and an ICRA 2025 demo reel that includes short sprints and boxing‑style motions :

Event demos, not deployments

Unitree told CNET that the running shots happened in unmapped spaces and were remote‑controlled. That is normal practice across humanoid labs because robust mapping, perception and safe planning in new places are still hard. Treat these as mechanics/control showcases rather than proofs of real‑world readiness. Strong demos still matter: they indicate actuator headroom, controller quality, and thermal limits under dynamic loads.

What looks improved in the gait

Stride timing appears more even, with quicker push‑off and less torso yaw. Arm swing is closer to human‑like phasing, which helps balance at speed. Foot placement looks surer on small irregularities, suggesting better state estimation and contact control. Combined, these changes point to improved whole‑body control and tuning, valuable foundations for future autonomy.

Unitree G1 price and who it is for

Unitree lists G1 at $16,000 on its official store, unusually low for a biped research platform (Store: https://shop.unitree.com/products/unitree-g1). The product page highlights imitation learning, reinforcement learning, dexterous hands, and wide joint ranges (Product page: https://www.unitree.com/g1). This price‑to‑capability mix targets labs, startups and advanced hobbyists who want to teach behaviors and iterate on policies rather than buy a turnkey worker.

Capabilities and claimed specs

Public spec roundups cite a top speed around 2 m/s and mass near 35 kg (example roundup: https://www.originofbots.com/robot/unitree-g1-by-unitree-robotics-details-specifications-rating). Treat third‑party tables as indicative until Unitree posts a locked spec sheet. For practical planning, focus on: continuous vs burst speed, battery swap/charge time, joint torque at key DOFs, and fall‑recovery behavior.

Where teleoperation still appears

When a clip notes that the site was not LiDAR‑mapped, assume teleoperation or tight supervision. Look for on‑screen overlays (localization maps, obstacle labels) or explicit “autonomous mode” callouts to verify autonomy. A good test is disturbance handling: minor shoves, unexpected obstacles, or route changes without a cut. If those are absent, the scene is likely curated.

H1 dancing robots and “whole‑body” control

At China’s Spring Festival Gala, Unitree’s taller H1 robots danced in sync with human performers. Reports attribute this to AI‑driven whole‑body motion control, which coordinates arms, legs and torso to track choreography while staying balanced (Business Insider coverage: https://www.businessinsider.com/china-spring-festival-robots-showing-off-humanoid-progress-2025-1). It’s a solid validation of timing, latency management and repeatability across multiple units.

What whole‑body control means in practice

In practice, whole‑body controllers solve for center‑of‑mass, foot placement and joint limits together, allowing expressive upper‑body motion without tripping the base. This improves both running (stability at speed) and dancing (style while balanced). It is not a substitute for perception, planning or manipulation; those layers decide where to go and what to do.

Multi‑robot sync vs real‑world surprises

Keeping many robots on‑beat shows tight clock sync, low actuation variance, and robust trajectory tracking. Real environments add non‑periodic surprises: people walking unpredictably, occlusions, slippery floors, and tasks requiring contact‑rich manipulation. Progress there will show up as longer uncut takes, visible perception stacks, and clear success‑rate metrics.

How G1 compares with other humanoids today

Most teams can now show agile motion. The harder step is a robot that navigates and manipulates without a script, safely, for hours, around people. Watch for: (1) clips in unmapped places still labeled autonomous; (2) task success rates over many trials; (3) fall recovery and safety documentation.

Agility Digit’s workcell strategy

Agility’s Digit emphasizes bounded workcells for logistics with dock charging and a pendant for moves between cells—a near‑term path that trades general navigation for reliability (Overview: https://www.agilityrobotics.com/solution). It’s a useful contrast with stage‑style agility demos: fewer hero moments, more repeatable throughput.

Sources & related information

CNET – Unitree’s G1 Humanoid Robots Shown Running in New Video – 2025

CNET reports that Unitree showed G1 running and H1 dancing, and adds that the running demos were teleoperated in unmapped spaces; these claims come directly from Unitree’s notes to the outlet (CNET report by Jesse Orrall).

Unitree – Unitree G1 – Product page – 2025

The official G1 page outlines the platform’s goals (embodied AI with imitation learning), highlights dexterous hands, and lists high-level specs suitable for research users (Unitree G1 product page).

Unitree – Unitree G1 – Price and purchase page – 2025

The official store listing shows G1 at $16,000, positioning it as a comparatively low‑cost biped platform for labs and developers (Unitree shop listing).

YouTube – G1 outdoor run at USC/RSS 2025 – 2025

Event footage shows G1 jogging outdoors on campus, illustrating smoother gait and steadier footfalls than earlier clips in a live setting (USC/RSS 2025 run).

YouTube – Unitree G1 demo reel (ICRA 2025) – 2025

A show‑floor reel captures short sprints and boxing‑style motions, typical of agility‑focused demonstrations under supervision (ICRA 2025 demo reel).

Agility Robotics – Digit – Solution overview – 2025

Agility describes a workcell‑first deployment model with dock charging and supervised moves, a contrast with stage‑style agility demos and useful context for readers (Agility solution overview).

Help us improve this article

Add a clear point? Share it in comment.

Link or video not working? Post a new one that works in comment.

Fact wrong or out-of-date? Write the correct info in comment.

Your feedback helps keep the content accurate and valuable.

 

0 Comments

Submit a Comment

Your email address will not be published. Required fields are marked *

Recently added

Listening is the underrated skill that makes you a better leader instantly

We often think of great leaders as great talkers. We imagine them giving rousing speeches, setting a clear vision, and having an answer for everything. But a massive review of scientific research suggests we have it backward. The most effective way to improve your leadership isn’t to speak more; it is to listen better.

New data shows that listening is not just a “soft skill” for making friends – it is a hard driver of job performance and professional success.

144 studies confirm listening drives performance

A recent meta-analysis published in the Journal of Business and Psychology examined the link between listening and work outcomes. The researchers looked at data from 144 studies involving more than 155,000 people.

Their conclusion was clear: listening has a strong, positive effect on employee job performance.

Leaders who are perceived as good listeners do more than just make their employees feel warm and fuzzy. They actually get better results. The study found that listening improves the quality of relationships at work, which in turn boosts performance. When employees feel heard, they perform better. This dynamic helps leaders unlearn bias and lower conflict within teams.

As the researchers noted, the link between listening and positive job outcomes is “robust.” They suggest that listening is an underrated predictor of job performance – a simple cause of superior results that many organizations overlook.

Why we love to talk about ourselves

If listening is so effective, why is it so hard? Why do so many of us default to talking instead?

The answer lies in our biology. A study published in the Proceedings of the National Academy of Sciences (PNAS) found that talking about ourselves is inherently rewarding. In fact, humans devote about 30–40 percent of everyday speech to informing others about their own subjective experiences – their thoughts, feelings, and opinions.

Using brain scans, researchers found that self-disclosure activates the mesolimbic dopamine system – the same brain regions associated with the pleasure we get from food, money, and sex. It feels good to talk about yourself.

The drive is so strong that people in the study were willing to give up money just to keep talking about themselves. When given a choice between answering questions about others for a higher payment or answering questions about themselves for a lower payment, participants voluntarily gave up between 17 and 25 percent of their potential earnings to talk about their own views.

We are wired to broadcast. To lead effectively, you have to fight that wiring.

The power of follow-up questions

You can become a better listener instantly by changing how you ask questions. It is not enough to just stay silent; you need to show you are engaged.

Research published in the Journal of Personality and Social Psychology found that the specific type of question you ask matters. The study showed that asking follow-up questions – questions that ask for more detail on what the other person just said – dramatically increases how likable you appear.

When you ask a follow-up question, you prove you were listening. You signal validation, care, and understanding. This simple habit makes you more persuasive and influential because, as other research in Frontiers in Psychology shows, likable people are better at influencing those around them.

Asking follow-up questions and recalling small details are among seven habits that mark an exceptional listener, and this research confirms it is a key tool for leaders.

Feeling known leads to feeling supported

Listening does more than build rapport; it meets a fundamental human need.

A study in the Journal of Personality and Social Psychology found that employees feel less objectified when their boss knows them as people, rather than just as workers or numbers. Furthermore, research linked to the Journal of Experimental Psychology suggests that “feeling known” is a necessary precursor to “feeling supported.”

You cannot support an employee you do not know. You cannot help them reach their career goals if you never asked what those goals were. You cannot solve their roadblocks if you never listened to what those roadblocks are.

What you can do about it

To become a better leader today, flip the ratio of your conversations.

  • Talk less. Recognize that your brain wants the dopamine hit of talking about yourself. Resist it.
  • Ask for their story, not yours. Instead of telling your team about your weekend or your problems, ask about theirs.
  • Use the follow-up rule. When an employee answers, do not just nod and move on. Restate what they said or ask one follow-up question based on what they just said.
  • Listen to learn. You already know what you know. The only way to learn something new is to listen to what others know.

Mastering conversation: how active listening keeps dialogue engaging is a skill you can practice in every interaction, whether with a colleague, a client, or a friend.

Sources & related information

Journal of Business and Psychology – The Power of Listening at Work – 2023

A meta-analysis of 144 studies involving 155,000 observations found that perceived listening is strongly correlated with improved job performance and relationship quality.

Proceedings of the National Academy of Sciences – Disclosing information about the self is intrinsically rewarding – 2012

Neuroimaging research shows that self-disclosure activates the brain’s reward systems, motivating people to talk about themselves even at a financial cost.

Journal of Personality and Social Psychology – It Doesn’t Hurt to Ask – 2017

A series of studies demonstrates that asking follow-up questions increases interpersonal liking by signaling responsiveness and listening.

The Pratfall Effect: why making mistakes can make you more likable

Perfection is often overrated. While we strive to be flawless in job interviews or first dates, psychology suggests that being too perfect can actually push people away. A small blunder, like tripping or spilling a drink, might do more for your popularity than a flawless performance. This phenomenon is known as the Pratfall Effect.

What is the Pratfall Effect?

The Pratfall Effect is a psychological principle that states that a person’s likability increases when they make a clumsy mistake, but only if that person is already perceived as competent.

Social psychologist Elliot Aronson first identified this effect in 1966. He wanted to test how mistakes influence attraction. In his famous experiment, he asked male college students to listen to tape recordings of people answering quiz questions.

The participants heard one of two main scenarios:

  1. The Superior Person: This person answered 92% of the questions correctly. They sounded confident and knowledgeable.
  2. The Average Person: This person answered only 30% of the questions correctly.

Aronson then added a twist. In some recordings, the “Superior Person” commits a blunder at the end: they are heard spilling a cup of coffee and reacting to the mess.

The results were clear. The students rated the Superior Person who spilled the coffee as the most likable of all. The blunder made the highly competent person seem more human and approachable.

The catch: competence is key

There is a crucial condition to this effect. A mistake only helps you if you have already established your competence.

In Aronson’s experiment, when the “Average Person” (who missed most quiz questions) spilled the coffee, their likability rating dropped even further.

  • If you are competent: A mistake humanizes you. It breaks the “too good to be true” barrier and prevents others from feeling threatened by your perfection.
  • If you are incompetent: A mistake just reinforces the idea that you are not capable. It acts as proof of inadequacy.

This distinction is vital. You cannot simply be clumsy and expect to be popular. You must first demonstrate that you are good at what you do. The blunder acts as a softener for your competence, not a substitute for it.

Real-world examples: from Jennifer Lawrence to brands

We see the Pratfall Effect in action in celebrity culture and marketing.

The relatable celebrity

Jennifer Lawrence is often cited as a modern example. Her frequent trips on the red carpet or candid, unpolished interviews often endear her to the public. Because she is an Oscar-winning, highly successful actress (high competence), these slips make her seem “down to earth” rather than clumsy.

The honest brand

Marketing experts use a similar concept known as the “blemishing effect.” When a brand admits a small flaw, consumers often trust it more. For example, Guinness: the beer brand famously turned a negative – the long time it takes to pour a pint – into a legendary slogan: “Good things come to those who wait.”

Why perfectionism harms connection

The Pratfall Effect challenges the idea that we must hide our flaws to be accepted. In social situations, perfection creates distance. We often struggle to connect with someone who seems to have no weaknesses because we cannot relate to them. This relates to understanding conversational biases to become more likable, where showing genuine engagement often matters more than saying the perfect thing.

When a competent person slips up, it levels the playing field. It signals vulnerability. This vulnerability fosters trust and signals that the person is authentic, not a curated persona.

What you can do about it

You do not need to stage accidents or spill coffee on purpose. However, you can change how you react to your own errors.

  • Don’t hide every flaw: If you are good at your job, admitting a small error or a gap in knowledge can make you more approachable to your team.
  • Own your blunders: When you trip or misspeak, laugh it off. Trying to cover it up often looks worse than the mistake itself.
  • Build competence first: Remember that this effect relies on a foundation of skill. Focus on being capable and reliable first.
  • Accept imperfection in others: Just as your mistakes humanize you, seeing others stumble is a reminder that everyone is human. This perspective can help reduce judgment and social anxiety.

Sources & related information

Elliot Aronson – The Effect of a Pratfall on Increasing Interpersonal Attractiveness – 1966

The original study published in Psychonomic Science where Aronson and his colleagues demonstrated that a blunder increases the attractiveness of a superior person but decreases the attractiveness of a mediocre person.

The Guardian (ZenithOptimedia) – The Pratfall effect and why brands should flaunt their flaws – 2015

An analysis of how brands like Guinness and VW use the Pratfall Effect to build trust by admitting minor weaknesses, making their core claims more believable.

Journal of Consumer Research – The blemishing effect – 2012

Research showing that under certain processing conditions, a small amount of negative information can actually enhance the positive impression of a product.

Endmyopia claims to reverse nearsightedness naturally (but science remains skeptical)

Imagine never needing your glasses again. No surgery, no contacts, just… fixing your eyes yourself. That’s the big promise of Endmyopia, a popular online method created by Jake Steiner. He claims you can reverse nearsightedness (myopia) just by changing your habits.

It sounds awesome, right? But before you throw away your glasses, you need to know that most eye doctors and scientists say it’s not that simple. Here is the lowdown on what this method is, why people try it, and why the medical consensus says it probably won’t work like you think.

The big claim: “Your glasses are the problem”

Endmyopia is based on a simple idea: your eyes aren’t broken; they are just reacting to your environment.

It starts with a muscle cramp

The theory goes like this: when you spend hours staring at your phone or laptop, a focusing muscle inside your eye gets tired and cramps up. This is called pseudo-myopia. At first, your vision is only blurry because of this cramp.

Then your eye grows longer

The controversial part is what happens next. Steiner says that when you wear glasses to fix that blur, your eye physically grows longer to “adapt” to the lenses. A longer eyeball is what causes true nearsightedness. Basically, the method claims your glasses trap you in a cycle that makes your vision worse.

The “fix”: training your eyes

To reverse this, Endmyopia tells you to do two things:

  1. Use weaker glasses: Instead of your full prescription, you wear weaker glasses for close-up work (like homework or gaming) to stop the eye strain.
  2. Practice “Active Focus”: This is a mental trick. You look at something far away that is slightly blurry (like a street sign) and try hard to make it clear just by focusing. The idea is that this effort forces your eyeball to shrink back to its normal size.

What science says

Here is the problem: Eye doctors agree that once your eyeball grows too long, it usually stays that way. It’s like your height – once you grow tall, you don’t shrink back down just because you want to.

Your eyeballs aren’t like muscles

You can train a muscle to get bigger, but you can’t really train an eyeball to get shorter. While atropine drops or special contact lenses can slow down eye growth in kids, there is no scientific proof that you can reverse it significantly once it’s happened.

Wearing weak glasses might backfire

Trying to fix your eyes by wearing weaker glasses can actually make things worse. A famous study (Chung et al., 2002) showed that under-correcting vision (wearing glasses that are too weak) made kids’ eyes grow faster, not slower. Blurry vision seems to signal the eye to keep growing, which is the exact opposite of what you want.

Why do some people swear it works?

If science says it doesn’t work, why are there so many success stories online? Read on Reddit: I was able to effectively fully cure myopia with my own methodology of eye exercises and discussions about do eye exercise really work?.

Brain training vs. Eye shrinking

When you practice looking at blurry things, your brain gets smarter at guessing what it’s seeing. This is called blur adaptation. You might be able to read a sign further away, not because your eyes are fixed, but because your brain is better at decoding the fuzzy image. You are “seeing” better, but your nearsightedness hasn’t actually disappeared.

Is there any hope?

Interestingly, some new research on red light therapy shows that specific light treatments might slightly shorten the eye.

Is it worth trying?

Trying to fix your eyes this way takes a huge amount of time – years of daily practice. Walking around (or driving!) with blurry vision can be dangerous.

Sources & related information

Study: Weak glasses make eyes worse (2002)

A major study showed that giving kids weaker glasses actually made their nearsightedness get worse faster.

Experts: Can you reverse myopia?

Eye doctors explain that while eye spasms can be fixed, the actual shape of a nearsighted eye is permanent.

Endmyopia Website

The source of the “active focus” method and the theory that glasses are to blame.

Red Light Therapy Study (2022)

A study showing that a specific type of red light therapy could shrink the eye slightly, proving some change is possible.