The Unitree Robotics LAFAN1 Retargeting Dataset visualized using Foxglove. ?? We had a little extra fun with this one—adding a feature to give the grid a fill color and adjusting another grid to shine down on the G1 dancing on the grass ??. Playfulness aside, this dataset employs some fascinating techniques. The retargeting process for this dataset uses numerical optimization techniques based on interaction mesh and inverse kinematics (IK). This approach ensures that retargeted motions adhere to end-effector pose constraints and joint position/velocity limits, effectively minimizing issues like foot slippage. However, this process focuses solely on kinematic constraints, excluding dynamic constraints and actuator limitations. As a result, while the dataset provides optimized motion trajectories, real robots may not execute these trajectories flawlessly due to unmodeled dynamic factors. But since Foxglove has a G1, we can always put it to the test, right? The LAFAN1 Retargeting Dataset is a valuable resource for researchers and developers working to improve humanoid robot motion. By providing retargeted motion data tailored to specific robot models, it helps advance the development of more natural and fluid robotic movements. Thanks for the dataset Unitree Robotics. Keep joy coming! Link to check out the dataset directly in Foxglove in the comments.
Foxglove
软件开发
San Francisco,CA 13,238 位关注者
Visualize, debug, and manage multimodal data in one purpose-built platform for robotics and embodied AI development.
关于我们
Foxglove's interactive visualization and data management capabilities empowers robotic developers to understand how their robots sense, think, and act in dynamic and unpredictable environments. All with the performance and scalability needed to create autonomy and build better robots, faster.
- 网站
-
https://foxglove.dev
Foxglove的外部链接
- 所属行业
- 软件开发
- 规模
- 51-200 人
- 总部
- San Francisco,CA
- 类型
- 私人持股
- 创立
- 2021
- 领域
- Multimodal Data Visualization、Multimodal Data Management、Robotics Development和Autonomous Robotics Development
产品
Foxglove
数据可视化软件
Understand how your robots sense, think, and act through Foxglove's multimodal data visualization and management platform. * Connect directly to your robot or inspect pre-recorded data. * Use interactive visualizations in customizable layouts to quickly understand what your robot is doing. * Overlay 2D and 3D images with lidar, point cloud, and other sensor data incl. annotations and bounding boxes to see what your robot sees. * Display and control robot models in interactive 3D scenes to understand how your robots move through the world. * Analyze everything as you travel through your robots mission and journey with Timeline and events. * Build bespoke panels, convert custom messages, and alias topic names to support your team’s unique development workflows. * Use Foxglove offline, store your data at the edge, then in your cloud or ours —no matter what your connectivity or data constraints are.
地点
-
主要
US,CA,San Francisco
Foxglove员工
动态
-
Foxglove转发了
FST Lisboa is proud to convey our deep appreciation to the outstanding team at Foxglove for their unwavering support. Foxglove’s innovative approach to enhancing the robotics development experience has been a perfect match for our objectives at FST Lisboa. Their advanced software solutions have had a tremendous impact, streamlining our workflow and helping us reach key milestones much faster. We are sincerely thankful to Foxglove for playing an essential role in the creation of our prototype and for their commitment to driving progress in the robotics sector. Their support has been a critical factor in our success.
-
-
?? Foxglove v2.23 is now available ?? Highlights ?? You can now preview topics before importing with the Foxglove Agent. The Foxglove Agent tracks metadata for each MCAP recording, including available topics, and forwards this information to Foxglove Cloud. This topic-level data is now accessible directly in Foxglove, allowing you to inspect and verify topics before importing recordings, ensuring a more efficient and informed data workflow. Plus ? The Plot panel now allows you to lock the x and y axis scales to a 1:1 ratio. You can enable this setting through the “Axis scales” dropdown in the panel settings, ensuring proportional data visualization for precise analysis. ? UI improvements – Hidden tooltips while dragging panels, better vertical scrollbar in State Transitions panel, and pretty-formatted timestamps in the Table panel. ? New Image panel support – View both original and converted messages with message converters. ? Recording filters – Filter recordings to show only those not linked to a device. ? Performance updates – Lower network usage for layout sync, faster H.264 video decoding, and optimized User Scripts loading. ? New API option – /data/stream now supports compressionFormat for MCAP chunk compression. Download and changelog links in the comments ??
-
Foxglove转发了
In person time is the key to successful remote work. What? At Foxglove we’re fully remote, but hold week-long offsites twice a year. It’s a great opportunity to build friendships, hold ad-hoc discussions, and brainstorm future product direction. Feeling very energized after spending the week with this incredible team!
-
-
Evelyn "Fifi" Yeung, Reframe Systems' founding engineer, is passionate about building scalable and intelligent robotic systems to unlock climate-friendly housing as a basic human right. Fifi and some of the Reframe team were at #Actuate2024 last year. Check out what they had to say as we get ready to announce #Actuate2025!
-
Foxglove转发了
The Unitree Robotics LAFAN1 Retargeting Dataset visualized using Foxglove. ?? We had a little extra fun with this one—adding a feature to give the grid a fill color and adjusting another grid to shine down on the G1 dancing on the grass ??. Playfulness aside, this dataset employs some fascinating techniques. The retargeting process for this dataset uses numerical optimization techniques based on interaction mesh and inverse kinematics (IK). This approach ensures that retargeted motions adhere to end-effector pose constraints and joint position/velocity limits, effectively minimizing issues like foot slippage. However, this process focuses solely on kinematic constraints, excluding dynamic constraints and actuator limitations. As a result, while the dataset provides optimized motion trajectories, real robots may not execute these trajectories flawlessly due to unmodeled dynamic factors. But since Foxglove has a G1, we can always put it to the test, right? The LAFAN1 Retargeting Dataset is a valuable resource for researchers and developers working to improve humanoid robot motion. By providing retargeted motion data tailored to specific robot models, it helps advance the development of more natural and fluid robotic movements. Thanks for the dataset Unitree Robotics. Keep joy coming! Link to check out the dataset directly in Foxglove in the comments.
Visualize and debug your robots faster.
-
?? Spotlight: Técnico Solar Boat, a student-led team at Instituto Superior Técnico, is developing advanced solar and hydrogen-powered vessels to push the boundaries of sustainable maritime technology. Their latest autonomous prototype, S?o Pedro 01 (SP01), integrates stereo cameras, lidar, and AI-driven navigation. Using Foxglove, they visualize and debug real-time sensor data for improved autonomy. Their projects, including hydrofoil-equipped solar boats and hydrogen fuel cell vessels, demonstrate the future of clean maritime mobility. Check out the full spotlight article in the comments ??
-
-
Adagy Robotics was in the house at #Actuate2024. Check out what Co-founder and CEO Rosalind Shinkle had to say about Adagy and Actuate. Are you ready for #Actuate2025?