Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

#CVPR2024 Highlight🌟 SimXR: Real-Time Simulated Avatar from Head-Mounted Sensors Controlling simulated humanoids and estimate full-body pose using SLAM images and headset pose in real time. 🌐: 📜: 🧑🏻‍💻: Input: four SLAM views from Quest 2 headset and 6DoF headset pose. Output: full-body humanoid control signals. Leveraging the laws of... show more

Zhengyi “Zen” Luo

6,198 subscribers

36,413 views • 2 years ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

11 Comments

Zhengyi “Zen” Luo2 years ago

How to learn a photon to actuations full-body control policy? Distill from PHC! First, we create large-scale synthetic data with paired full-body motion and egocentric images. Then, train a control policy to consume these input streams and query PHC for pseudo-gt torques as supervision.

Zhengyi “Zen” Luo2 years ago

While we only train using synthetic data, SimXR generalizes to real-world data captures. Here is some synthetic data results: we randomize background and lighting at a per-frame level.

Zhengyi “Zen” Luo2 years ago

This pipeline is also applicable to AR headsets! Here is some results trained and tested on the Aria Digital Twin (ADT) dataset (two views).

Zhengyi “Zen” Luo2 years ago

All code and data will be released at Repo is a work in progress at the moment 🤫

Zhengyi “Zen” Luo2 years ago

Thanks to everyone on the team! @jinkuncao @Me_Rawal @awinkler_ Jing Huang @kkitani @xuweipeng000 Work done at Meta Reality Labs, Pittsburgh in collaboration with CMU.

Zhengyi “Zen” Luo2 years ago

Thoughts: learning a universal humanoid motion imitator that has scaled to a large motion dataset like AMASS really pays off. For tasks like animation, pose estimation, and teleoperation, either in simulation only or real-world (cue a strong imitator can already be used for many tasks (using kinematic pose as the intermediate representation). It is also the gateway for acquiring motor skills (cue where you can learn compact motion representations that will be universally applicable to downstream tasks. PULSE has been tremendously useful in my projects, which I will share soon 😁

Chen Tessler2 years ago

Nice! I need to hook up my quest to Isaac. This looks so much fun! 🤩

FutureMapper ᯅ2 years ago

How are you accessing the camera?

PowerBeatsVR3 years ago

Get ready for a full-body VR workout that’s fun, fast, and intuitive — Play PowerBeatsVR (Now on Meta Quest) 🔥

haareblond2 years ago

wow cool this might make upper or full body tracking possible for some diy projects. very impressive. :o

Zhengyi “Zen” Luo2 years ago

Meta’s IOBT will add support for that very soon 😇

Related Videos

‘CLONE’ – whole-body teleoperation of a humanoid. Intuitive control signals are captured by tracking the teleoperator’s head and hand poses using Apple Vision Pro. A Mixture-of-Experts policy takes the sparse input and synthesizes the corresponding whole-body humanoid pose.

‘CLONE’ – whole-body teleoperation of a humanoid. Intuitive control signals are captured by tracking the teleoperator’s head and hand poses using Apple Vision Pro. A Mixture-of-Experts policy takes the sparse input and synthesizes the corresponding whole-body humanoid pose.

The Humanoid Hub

38,155 views • 1 year ago

Estimating a full-body pose from only the headset and controllers is ambiguous. Environment information can help resolve this. In our #SIGGRAPH2023 paper the avatar is generated from only the 3 shown coordinate frames (no cameras) + height of the environment (green dots). (1/4)👇

Alexander Winkler

200,300 views • 3 years ago

WIP. The final piece of the puzzle. A dedicated tail rig, unlike the full body with this rig you can control and drag from the tail tip to pose for easier control.

WIP. The final piece of the puzzle. A dedicated tail rig, unlike the full body with this rig you can control and drag from the tail tip to pose for easier control.

snek3d

18,555 views • 8 months ago

Meet LA-Pose. Our latest model taking Wayve another step towards generalization at scale. LA-Pose employs large-scale self-supervised learning, building strong motion representations for 3D perception from 10.2 million unlabeled driving video snippets, unlike today's strongest approaches that often depend on expensive, carefully curated 3D supervision. With only a lightweight pose head and limited labelled data, LA-Pose achieves: 📷 State-of-the-art camera pose estimation 🌎 Strong zero-shot generalization across diverse driving scenarios 🏷️ Orders of magnitude less labelled data than fully supervised 3D approaches Our full blog post: Explore the full paper here:

Meet LA-Pose. Our latest model taking Wayve another step towards generalization at scale. LA-Pose employs large-scale self-supervised learning, building strong motion representations for 3D perception from 10.2 million unlabeled driving video snippets, unlike today's strongest approaches that often depend on expensive, carefully curated 3D supervision. With only a lightweight pose head and limited labelled data, LA-Pose achieves: 📷 State-of-the-art camera pose estimation 🌎 Strong zero-shot generalization across diverse driving scenarios 🏷️ Orders of magnitude less labelled data than fully supervised 3D approaches Our full blog post: Explore the full paper here:

Wayve

36,915 views • 4 months ago

What would a World Model look like if we start from a real embodied agent acting in the real world? It has to have: 1) A real, physically grounded and complex action space—not just abstract control signals. 2) Diverse, real-life scenarios and activities. Or in short: It has to be annoyingly complex—in both the action and vision space—to even get close to real life. We did an initial attempt: Whole-Body Conditioned Egocentric Video Prediction. In collaboration with @dans_t123 , Amir Bar, Yann LeCun , trevordarrell and Jitendra MALIK. (For more details, check: What we did is very simple: Predict Egocentric Video from human Actions (PEVA) - Given the past video and a future action represented by relative 3D body pose, PEVA predicts how the world looks next—from the first-person view. By conditioning on kinematic pose trajectories, structured by the joint hierarchy of the body, it learns how physical actions shape perception.

What would a World Model look like if we start from a real embodied agent acting in the real world? It has to have: 1) A real, physically grounded and complex action space—not just abstract control signals. 2) Diverse, real-life scenarios and activities. Or in short: It has to be annoyingly complex—in both the action and vision space—to even get close to real life. We did an initial attempt: Whole-Body Conditioned Egocentric Video Prediction. In collaboration with @dans_t123 , Amir Bar, Yann LeCun , trevordarrell and Jitendra MALIK. (For more details, check: What we did is very simple: Predict Egocentric Video from human Actions (PEVA) - Given the past video and a future action represented by relative 3D body pose, PEVA predicts how the world looks next—from the first-person view. By conditioning on kinematic pose trajectories, structured by the joint hierarchy of the body, it learns how physical actions shape perception.

Yutong Bai

179,519 views • 1 year ago

🎛️#Live3D #Live2D #Live2DWIP🎛️ Aside from the 3D palm made in Live2D, today I’d like to show a pendulum-based inverse kinematics and pose system. The puppet starts in a T-pose, and when I move the parameters, they pass through a logic gate simulated by physics pendulums. This distributes the pose into the correct XYZ angles for the upper arm, forearm, and palm. That’s how it creates both the relaxed and raised arm movements seen in the video. The IK system works in a similar way. It maps the captured XY position of the palm to the arm’s XY rotation, making it possible to “move the hand” in space. These systems have been floating in my head for a long time, and today I finally confirmed they’re feasible. I plan to apply this further by adding more preset poses and new control methods to the puppet.

🎛️#Live3D #Live2D #Live2DWIP🎛️ Aside from the 3D palm made in Live2D, today I’d like to show a pendulum-based inverse kinematics and pose system. The puppet starts in a T-pose, and when I move the parameters, they pass through a logic gate simulated by physics pendulums. This distributes the pose into the correct XYZ angles for the upper arm, forearm, and palm. That’s how it creates both the relaxed and raised arm movements seen in the video. The IK system works in a similar way. It maps the captured XY position of the palm to the arm’s XY rotation, making it possible to “move the hand” in space. These systems have been floating in my head for a long time, and today I finally confirmed they’re feasible. I plan to apply this further by adding more preset poses and new control methods to the puppet.

📐Hephaestus📏Live2D匠人魂

10,984 views • 1 year ago

How do you give a humanoid the general motion capability? Not just single motions, but all motion? Introducing SONIC, our new work on supersizing motion tracking for natural humanoid control. We argue that motion tracking is the scalable foundation task for humanoids. So we "supersized" it: 9k+ GPU hours and 100M+ motion frames. But tracking alone is not enough; we show how to make a useful control system out of it: - Universal Kinematic Planner: Enables game-like gamepad control and high-level teleoperation, just like controlling a character in a game. - VR Full-Body Teleop: Direct, real-time whole-body control by a human wearing a VR headset. - VR Keypoint Teleop: Control the upper body (hands/head) while our planner handles robust locomotion automatically. - VLA Integration: We connect this motion tracker to autonomous Visual-Language-Action (VLA) models for autonomous task execution! We use a Universal Token Space to UNIFY this command space, turning our robust tracker into a general-purpose, programmable humanoid brain. This is the generalist "System 1" for humanoids. 🚀 Project: #Humanoids #Robotics #AI #FoundationModels #NVIDIAResearch 🧠🔥

How do you give a humanoid the general motion capability? Not just single motions, but all motion? Introducing SONIC, our new work on supersizing motion tracking for natural humanoid control. We argue that motion tracking is the scalable foundation task for humanoids. So we "supersized" it: 9k+ GPU hours and 100M+ motion frames. But tracking alone is not enough; we show how to make a useful control system out of it: - Universal Kinematic Planner: Enables game-like gamepad control and high-level teleoperation, just like controlling a character in a game. - VR Full-Body Teleop: Direct, real-time whole-body control by a human wearing a VR headset. - VR Keypoint Teleop: Control the upper body (hands/head) while our planner handles robust locomotion automatically. - VLA Integration: We connect this motion tracker to autonomous Visual-Language-Action (VLA) models for autonomous task execution! We use a Universal Token Space to UNIFY this command space, turning our robust tracker into a general-purpose, programmable humanoid brain. This is the generalist "System 1" for humanoids. 🚀 Project: #Humanoids #Robotics #AI #FoundationModels #NVIDIAResearch 🧠🔥

Zhengyi “Zen” Luo

62,515 views • 10 months ago

This man was in the gym flexing his massive physique during a posing session. He’s shirtless and oiled up, hitting every classic pose from double biceps to side chest and turning to show off his back while other lifters stand around watching and filming. He looks completely confident and ripped the whole time. I give him props for building a body like that, would you ever get on stage and pose in front of a crowd like him?

This man was in the gym flexing his massive physique during a posing session. He’s shirtless and oiled up, hitting every classic pose from double biceps to side chest and turning to show off his back while other lifters stand around watching and filming. He looks completely confident and ripped the whole time. I give him props for building a body like that, would you ever get on stage and pose in front of a crowd like him?

R͓̽Y͓̽a͓̽n͓̽C͓̽e͓̽y͓̽

3,435,931 views • 5 months ago

The largest fully annotated open egocentric human dataset. Today we're open-sourcing EgoSuite-Open100K with Hugging Face : 100,000 hours of egocentric human data, free for anyone to use. Physical AI has a scaling law of its own now, and its input is human data. That part is settled. What the field still needs is a shared foundation to build on. 15,000+ tasks. 15,000+ real scenes, from factory floors to retail backrooms. Hand pose, body pose and subtask-level semantics, with wrist camera coverage on part of the set. Licensed for commercial training, not research-only. We build the data infrastructure for Physical AI. Open100K is the first public layer of it. The first 10,000 hours are live now, with the rest rolling out in stages.

The largest fully annotated open egocentric human dataset. Today we're open-sourcing EgoSuite-Open100K with Hugging Face : 100,000 hours of egocentric human data, free for anyone to use. Physical AI has a scaling law of its own now, and its input is human data. That part is settled. What the field still needs is a shared foundation to build on. 15,000+ tasks. 15,000+ real scenes, from factory floors to retail backrooms. Hand pose, body pose and subtask-level semantics, with wrist camera coverage on part of the set. Licensed for commercial training, not research-only. We build the data infrastructure for Physical AI. Open100K is the first public layer of it. The first 10,000 hours are live now, with the rest rolling out in stages.

Lightwheel

73,314 views • 20 days ago

So cool...a humanoid just drove a go-kart for real. A Unitree G1 lowers itself into the tight cockpit, hands on the wheel, feet on the pedals, then runs the track， even drifts. It shows constrained whole-body loco-manipulation, mobile pick-and-place, and hand-eye-foot coordination in one sequence. This comes from China’s Symbiosis Robotics and their Direct Perception Control (DPC) model. It skips the usual high-level goals + tracker stack, mapping vision, language, body state and feedback straight to joint and hand targets. The focus is whole-body coordination for loco-manipulation. Interestingly, data diversity...Training used 15,010 hours of joint-aligned data-- 6,781h human egocentric, 4,024h armed robots, 3,660h wheeled humanoids, 545h bipedal. Teleop, head-mounted and egocentric sources were all converted into time-aligned joint trajectories across embodiments. A clean test that packs perception, balance and continuous whole-body control into one task chain ,and leaves room for other real-world scenarios.

So cool...a humanoid just drove a go-kart for real. A Unitree G1 lowers itself into the tight cockpit, hands on the wheel, feet on the pedals, then runs the track， even drifts. It shows constrained whole-body loco-manipulation, mobile pick-and-place, and hand-eye-foot coordination in one sequence. This comes from China’s Symbiosis Robotics and their Direct Perception Control (DPC) model. It skips the usual high-level goals + tracker stack, mapping vision, language, body state and feedback straight to joint and hand targets. The focus is whole-body coordination for loco-manipulation. Interestingly, data diversity...Training used 15,010 hours of joint-aligned data-- 6,781h human egocentric, 4,024h armed robots, 3,660h wheeled humanoids, 545h bipedal. Teleop, head-mounted and egocentric sources were all converted into time-aligned joint trajectories across embodiments. A clean test that packs perception, balance and continuous whole-body control into one task chain ,and leaves room for other real-world scenarios.

CyberRobo

76,039 views • 24 days ago

🤔 Ever wondered if simulation-based animation/avatar learnings can be applied to real humanoid in real-time? 🤖 Introducing H2O (Human2HumanOid): - 🧠 An RL-based human-to-humanoid real-time whole-body teleoperation framework - 💃 Scalable retargeting and training using large human motion dataset - 🎥 With just an RGB camera, everyone can teleoperate a full-sized humanoid to perform actions like pick and place, walking, kicking, boxing, etc - 💡Unleash the potential of humanoids with human cognitive skills and adaptability 🔗: 📄: 🎬: H2O proposes: - A scalable retargeting framework for obtaining large-scale humanoid motion dataset, intelligently filtering out infeasible motion for the humanoid embodiment. - We train a full-body motion imitator (similar to PHC) and deploy to the real world zero-shot. - Using this framework, we enable real-time teleoperation of a humanoid via a human operator and webcam, performing skills such as pick and place, kicking, walking strollers, etc. Team: Tairan He Zhengyi “Zen” Luo Wenli Xiao @ChongZitaZhang Kris Kitani Changliu Liu Guanya Shi

Zhengyi “Zen” Luo

47,323 views • 2 years ago

This seemingly obvious prediction didn't take long to become reality. MASt3R-SLAM: Real-Time Dense SLAM with 3D Reconstruction Priors Contributions: • The first real-time SLAM system using the two-view 3D reconstruction prior MASt3R [20] as a foundation. • Efficient techniques for pointmap matching, tracking and local fusion, graph construction and loop closure, and second-order global optimization. • A state-of-the-art dense SLAM system capable of handling generic, time-varying camera models. Abstract: We present a real-time monocular dense SLAM system, designed from the ground up using MASt3R, a two-view 3D reconstruction and matching prior. Equipped with this strong prior, our system remains robust on in-the-wild video sequences, making no assumptions on a fixed or parametric camera model beyond a unique camera center. Key features include: - Efficient methods for pointmap matching, camera tracking, and local fusion - Graph construction and loop closure - Second-order global optimization With known calibration, a simple modification achieves state-of-the-art performance across various benchmarks. Altogether, we propose a plug-and-play monocular SLAM system capable of producing globally-consistent poses and dense geometry while operating at 15 FPS.

This seemingly obvious prediction didn't take long to become reality. MASt3R-SLAM: Real-Time Dense SLAM with 3D Reconstruction Priors Contributions: • The first real-time SLAM system using the two-view 3D reconstruction prior MASt3R [20] as a foundation. • Efficient techniques for pointmap matching, tracking and local fusion, graph construction and loop closure, and second-order global optimization. • A state-of-the-art dense SLAM system capable of handling generic, time-varying camera models. Abstract: We present a real-time monocular dense SLAM system, designed from the ground up using MASt3R, a two-view 3D reconstruction and matching prior. Equipped with this strong prior, our system remains robust on in-the-wild video sequences, making no assumptions on a fixed or parametric camera model beyond a unique camera center. Key features include: - Efficient methods for pointmap matching, camera tracking, and local fusion - Graph construction and loop closure - Second-order global optimization With known calibration, a simple modification achieves state-of-the-art performance across various benchmarks. Altogether, we propose a plug-and-play monocular SLAM system capable of producing globally-consistent poses and dense geometry while operating at 15 FPS.

MrNeRF

29,974 views • 1 year ago

an uncomfortable amount of time has passed since the last update, hasn't it, huh! aside from various background work, I made a full-blown almanac with stats and even a real-time simulated showcase for babies and monsters, as the game got a little too heavy with new weird stuff..

an uncomfortable amount of time has passed since the last update, hasn't it, huh! aside from various background work, I made a full-blown almanac with stats and even a real-time simulated showcase for babies and monsters, as the game got a little too heavy with new weird stuff..

toncho_

49,144 views • 1 year ago

ViPE: Video Pose Engine for 3D Geometric Perception Contributions: • A robust and efficient framework, ViPE, for estimating camera parameters and dense depth from diverse, in-the-wild videos. • A system design that integrates the strengths of classical SLAM (efficiency, scalability) and learned models (robustness), with key improvements in efficiency, dynamic object handling, and depth quality over prior work. • A large-scale dataset of annotated videos, created using ViPE, to facilitate future research in 3D computer vision.

ViPE: Video Pose Engine for 3D Geometric Perception Contributions: • A robust and efficient framework, ViPE, for estimating camera parameters and dense depth from diverse, in-the-wild videos. • A system design that integrates the strengths of classical SLAM (efficiency, scalability) and learned models (robustness), with key improvements in efficiency, dynamic object handling, and depth quality over prior work. • A large-scale dataset of annotated videos, created using ViPE, to facilitate future research in 3D computer vision.

MrNeRF

42,641 views • 1 year ago

Physics-based Motion Retargeting from Sparse Inputs paper page: Avatars are important to create interactive and immersive experiences in virtual worlds. One challenge in animating these characters to mimic a user's motion is that commercial AR/VR products consist only of a headset and controllers, providing very limited sensor data of the user's pose. Another challenge is that an avatar might have a different skeleton structure than a human and the mapping between them is unclear. In this work we address both of these challenges. We introduce a method to retarget motions in real-time from sparse human sensor data to characters of various morphologies. Our method uses reinforcement learning to train a policy to control characters in a physics simulator. We only require human motion capture data for training, without relying on artist-generated animations for each avatar. This allows us to use large motion capture datasets to train general policies that can track unseen users from real and sparse data in real-time. We demonstrate the feasibility of our approach on three characters with different skeleton structure: a dinosaur, a mouse-like creature and a human. We show that the avatar poses often match the user surprisingly well, despite having no sensor information of the lower body available. We discuss and ablate the important components in our framework, specifically the kinematic retargeting step, the imitation, contact and action reward as well as our asymmetric actor-critic observations. We further explore the robustness of our method in a variety of settings including unbalancing, dancing and sports motions.

AK

106,527 views • 3 years ago

SqueezeMe: Efficient Gaussian Avatars for VR TL;DR: Three of these Gaussian Splatting avatars can be run at 72 frames per second. It runs locally on a Meta Quest 3 VR headset. Abstract (excerpt): While previous methods require a desktop GPU for real-time inference of a single avatar, we aim to squeeze multiple Gaussian avatars onto a portable virtual reality headset with real-time drivable inference. We begin by training a previous work, Animatable Gaussians, on a high-quality dataset captured with 512 cameras. The Gaussians are animated by controlling a base set of Gaussians with linear blend skinning (LBS) motion, and then further adjusting them with a neural network decoder to correct their appearance. When deploying the model on a Meta Quest 3 VR headset, we find two major computational bottlenecks: the decoder and the rendering. To accelerate the decoder, we train the Gaussians in UV-space instead of pixel-space and distill the decoder to a single neural network layer. Further, we discover that neighborhoods of Gaussians can share a single corrective from the decoder, providing an additional speedup. To accelerate the rendering, we develop a custom pipeline in Vulkan that runs on the mobile GPU. Putting it all together, we run 3 Gaussian avatars concurrently at 72 FPS on a VR headset.

MrNeRF

27,104 views • 1 year ago

With Helix 02, Figure has introduced a new foundational layer for whole-body control that replaces 109k lines of hand-engineered C++ with a single neural prior for stable, natural motion. - Learned from 1,000+ hours of human motion data and sim-to-real RL across 200,000+ parallel environments. - A 10M-parameter neural network that takes full-body joint state and base motion as input and outputs joint-level actuator commands at 1 kHz (1,000 times a second).

With Helix 02, Figure has introduced a new foundational layer for whole-body control that replaces 109k lines of hand-engineered C++ with a single neural prior for stable, natural motion. - Learned from 1,000+ hours of human motion data and sim-to-real RL across 200,000+ parallel environments. - A 10M-parameter neural network that takes full-body joint state and base motion as input and outputs joint-level actuator commands at 1 kHz (1,000 times a second).

The Humanoid Hub

28,004 views • 7 months ago

Meta presents Sapiens Foundation for Human Vision Models discuss: We present Sapiens, a family of models for four fundamental human-centric vision tasks - 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction. Our models natively support 1K high-resolution inference and are extremely easy to adapt for individual tasks by simply fine-tuning models pretrained on over 300 million in-the-wild human images. We observe that, given the same computational budget, self-supervised pretraining on a curated dataset of human images significantly boosts the performance for a diverse set of human-centric tasks. The resulting models exhibit remarkable generalization to in-the-wild data, even when labeled data is scarce or entirely synthetic. Our simple model design also brings scalability - model performance across tasks improves as we scale the number of parameters from 0.3 to 2 billion. Sapiens consistently surpasses existing baselines across various human-centric benchmarks. We achieve significant improvements over the prior state-of-the-art on Humans-5K (pose) by 7.6 mAP, Humans-2K (part-seg) by 17.1 mIoU, Hi4D (depth) by 22.4% relative RMSE, and THuman2 (normal) by 53.5% relative angular error.

Meta presents Sapiens Foundation for Human Vision Models discuss: We present Sapiens, a family of models for four fundamental human-centric vision tasks - 2D pose estimation, body-part segmentation, depth estimation, and surface normal prediction. Our models natively support 1K high-resolution inference and are extremely easy to adapt for individual tasks by simply fine-tuning models pretrained on over 300 million in-the-wild human images. We observe that, given the same computational budget, self-supervised pretraining on a curated dataset of human images significantly boosts the performance for a diverse set of human-centric tasks. The resulting models exhibit remarkable generalization to in-the-wild data, even when labeled data is scarce or entirely synthetic. Our simple model design also brings scalability - model performance across tasks improves as we scale the number of parameters from 0.3 to 2 billion. Sapiens consistently surpasses existing baselines across various human-centric benchmarks. We achieve significant improvements over the prior state-of-the-art on Humans-5K (pose) by 7.6 mAP, Humans-2K (part-seg) by 17.1 mIoU, Hi4D (depth) by 22.4% relative RMSE, and THuman2 (normal) by 53.5% relative angular error.

AK

151,614 views • 2 years ago

We have SLAM on the Robocap! 🎉 Visualized with Rerun Using NVIDIA AI Developer cuVSLAM for GPU-accelerated multicamera tracking. I basically wrote zero code myself and fully used Claude Code for this. It worked because I had so many existing examples to point to that it just wrote everything the way I would have. A few technical wins: 1. Used rattler-build from prefix.dev to package the compiled cuVSLAM CUDA binaries, which made it SUPER easy to use across repos. This also means it works on the DGX Spark (ARM64) out of the box. 2. Zero-setup experience: git clone && pixi run track-robocap auto-downloads a 100MB dataset from HuggingFace and tracks frames. 3. Real-time 3D visualization with trajectories, landmarks, pose graphs, and video playback in Rerun. Still visual-only (not visual-inertial yet), and loop closure needs some debugging. Next steps are getting this into a Gradio interface, then into daggr, and extending it to work with other datasets from exoego-forge. The last piece I'm excited about: Rerun's RRD files now support layers for incremental data. Planning to build pipelines that go from raw sensor data → slam -> human pose → depth estimation → etc. Repo here:

We have SLAM on the Robocap! 🎉 Visualized with Rerun Using NVIDIA AI Developer cuVSLAM for GPU-accelerated multicamera tracking. I basically wrote zero code myself and fully used Claude Code for this. It worked because I had so many existing examples to point to that it just wrote everything the way I would have. A few technical wins: 1. Used rattler-build from prefix.dev to package the compiled cuVSLAM CUDA binaries, which made it SUPER easy to use across repos. This also means it works on the DGX Spark (ARM64) out of the box. 2. Zero-setup experience: git clone && pixi run track-robocap auto-downloads a 100MB dataset from HuggingFace and tracks frames. 3. Real-time 3D visualization with trajectories, landmarks, pose graphs, and video playback in Rerun. Still visual-only (not visual-inertial yet), and loop closure needs some debugging. Next steps are getting this into a Gradio interface, then into daggr, and extending it to work with other datasets from exoego-forge. The last piece I'm excited about: Rerun's RRD files now support layers for incremental data. Planning to build pipelines that go from raw sensor data → slam -> human pose → depth estimation → etc. Repo here:

Pablo Vela

50,825 views • 6 months ago