Leixin Chang

Robotics PhD student at Zhejiang University

[Leixin Chang @leixinchang]

@NicholasEPfaff Great work! Curious about how do you make the physical params of simulated objects plausible? And how do you evaluate?

[Flexion @FlexionAI]

"A parcel with snacks has been delivered for Flexion. Retrieve it using the stairs and come up using the elevator. Then unpack it and place the items into the empty drawer on the shelf in the snack area."

One instruction. No human operator. Everything that follows is autonomous.

Today we're introducing Reflect v1.0, our robotics intelligence platform for long-horizon work. From a single natural-language command, the robot understands the task, navigates a multi-floor building, calls elevators, handles doors, uses tools to unpack a box, and puts the items away. The biggest shift in v1.0 is that we use reinforcement learning across every layer, from low-level control to high-level reasoning.

Long-horizon autonomy is unforgiving. The robot must recover on its own when things don't go to plan because in the real world, they never do. Combining reasoning, perception, physical execution and runtime robustness into a single mission-capable system is the foundation required to solve humanoid autonomy.

Our team is just getting started.

#HumanoidRobots #Flexion

[Leixin Chang @leixinchang]

@realyunlong From RL perspective, imitation learning bypasses the issue of how Tod define the task.

[Leixin Chang @leixinchang]

Fast adapt WBC policy from one robot to another. Like transferring sonic policy from G1 to other humanoids.

[Tao Yu @tao_robotics]

Introducing Any2Any: efficient cross-embodiment transfer for humanoid whole-body tracking.

With only 1% compute/data, we transfer Gear-Sonic from Unitree G1 to LimX Oli/Luna with fast convergence and strong tracking.

Check out our initial version on arXiv: 2605.23733

#sonic

[Leixin Chang @leixinchang]

@jsonbot117 Go for it! You've done great work!

[Weishuai Zeng @weishuaizeng]

Since we first put forward the concept of Behavior Foundation Model (BFM) last September, we have been exploring ways to fully unleash its potential.

Here comes our answer: Scaling Behavior Foundation Model for Humanoid Robots.

[Leixin Chang @leixinchang]

Agentic 3D asset creation for automatic sim scene generation + high-visual-fidelity simulator --> train the scene-specific robot policy in simulation as a warm start 🤔

[RuiningLi @RayLi234]

🚀 Introducing Articraft, a coding agent for articulated 3D asset creation.

Articraft writes code, executes it, receives validation feedback, and refines the result into simulation-ready 3D assets with parts, joints, and motion.

We’re also releasing Articraft-10K: 10,000+ articulated objects across 250 categories, unlocking large-scale interactive scenes for robotics simulation and physical AI.

🔗 Project page: https://t.co/FWutv61yx7
💻 Code: https://t.co/CpCYdBzMlv

[Andy Zeng @andyzengineer]

For the avid viewer -- there’s a brief moment when the robot loses it’s grip on the head of a ziptie, and so it decides to use the other hand to help readjust the grip for the pull.

It’s gnarly passing by our robots everyday, and catching these random glimpses of improvisational intelligence in action. Instant dopamine hit.

[Patrick Yin @patrickhyin]

We’re releasing OmniReset, a framework for training robot policies using large-scale RL and diverse resets for contact-rich, dexterous manipulation.

OmniReset pushes the frontier of robustness and dexterity, without any reward engineering or demonstrations.

Try the policies yourself in our interactive simulator! https://t.co/3hW3nYx2vD

(1/N 🧵)

[Junzhe (JJ) He @JayHe748646]

Will release something interesting in the coming months.

[Chenhao Li @breadli428]

Simplicity should be valued more. When a task can be solved equally with a simpler framework, one should not be blamed having “nothing new”.

Many unnecessary novelties are invented for the sake of novelty, while the effort of making simpler methods general is not appreciated.

[Physical Intelligence @physical_int]

We’ve developed a memory system for our models that provides both short-term visual memory and long-term semantic memory.

Our approach allows us to train robots to perform long and complex tasks, like cleaning up a kitchen or preparing a grilled cheese sandwich from scratch 👇

[Ruiqian Nai @RuiqianNai]

🤖 Can we demonstrate humanoid complex whole-body manipulation skills without a physical robot present?
Introducing HuMI: A portable, robot-free interface for learning diverse humanoid manipulation tasks.
📄 https://t.co/bKdgzgTnQb
🌐 https://t.co/K7nWFORg3b

[Michael Xu @mxu_cg]

Check out this repo if you want to play with this real-time character controller (need RTX 4090): https://t.co/yGlmoZiBvV

The dataset and models are here: https://t.co/mjaZYbNk4d

I also wrote a blog post about the autophagous data augmentation method: https://t.co/YCJUiygJEU

[Junzhe (JJ) He @JayHe748646]

Tracking is too limited. Now we do navigation-based parkour!

[Leixin Chang @leixinchang]

@jsonbot117 niubi

[Alper Canberk @alpercanbe]

Earlier today, we asked Memo to grasp whatever that’s in front of it. Here's a short thread of emergent behaviors 🧵

[Junzhe (JJ) He @JayHe748646]

@KehanWen170077 and I trained a heading controlled humanoid parkour policy that has some recovery ability from falling (notice the recovery behavior when climbing the second stage). Looks quite promising! Doing hardware experiments and wrapping up everything now. Stay tuned!

[Chaoyi Pan @ChaoyiPan]

Generative models (diffusion/flow) are taking over robotics 🤖. But do we really need to model the full action distribution to control a robot?

We suspected the success of Generative Control Policies (GCPs) might be "Much Ado About Noising."

We rigorously tested the myths. 🧵👇

[Jason Peng @xbpeng4]

MimicKit now supports #IsaacLab! After many years with IsaacGym, it's time to upgrade. MimicKit has a simple Engine API that allows you to easily swap between different simulator backends.

Which simulator would you like to see next?