Update
Update
Update
001
001
001
A sneak peek at our v0 in the pursuit of a SOTA robotics data collection platform.
A sneak peek at our v0 in the pursuit of a SOTA robotics data collection platform.
MOTION RETARGETING
MOTION RETARGETING
SPATIAL CONTEXT
SPATIAL CONTEXT
GRASP DETECTION
GRASP DETECTION
SCENE ANALYSIS
SCENE ANALYSIS
RAW INPUT VIDEO
RAW INPUT VIDEO
We’ve been heads-down tackling one of the biggest blockers for building robot foundation models: high-quality, scalable data. Our goal is to make it way easier to collect and unify data gathered via teleop, synthetic environments, and video.
First up—we’re excited to launch video2embodiment, a pipeline that turns everyday videos into training-ready robot data. It adds rich spatial context and metadata, and makes the whole process—from request to delivery—simple, streamlined, and fully servicified.
More updates are coming soon as we roll out support for other data collection methods. Stay tuned!
We’ve been heads-down tackling one of the biggest blockers for building robot foundation models: high-quality, scalable data. Our goal is to make it way easier to collect and unify data gathered via teleop, synthetic environments, and video.
First up—we’re excited to launch video2embodiment, a pipeline that turns everyday videos into training-ready robot data. It adds rich spatial context and metadata, and makes the whole process—from request to delivery—simple, streamlined, and fully servicified.
More updates are coming soon as we roll out support for other data collection methods. Stay tuned!
Scalable, Multimodal Data Collection
Scalable, Multimodal Data Collection
High-Fidelity Video + Depth
Capture rich visual data streams with synchronized RGB and depth for spatially aware training.
High-Fidelity Video + Depth
Capture rich visual data streams with synchronized RGB and depth for spatially aware training.
High-Fidelity Video + Depth
Capture rich visual data streams with synchronized RGB and depth for spatially aware training.
Precise Robot & Camera Trajectories
Accurate localization and full trajectory tracking for both robots and onboard cameras.
Precise Robot & Camera Trajectories
Accurate localization and full trajectory tracking for both robots and onboard cameras.
Precise Robot & Camera Trajectories
Accurate localization and full trajectory tracking for both robots and onboard cameras.
Audio for Narrative & Event Understanding
Capture ambient audio for labeling actions, detecting events, and enriching contextual learning.
Audio for Narrative & Event Understanding
Capture ambient audio for labeling actions, detecting events, and enriching contextual learning.
Audio for Narrative & Event Understanding
Capture ambient audio for labeling actions, detecting events, and enriching contextual learning.
WIP: Tactile & Wrist-Mounted Sensing
Integrating pressure sensors and wrist cameras to enhance manipulation understanding and fine-grained control.
WIP: Tactile & Wrist-Mounted Sensing
Integrating pressure sensors and wrist cameras to enhance manipulation understanding and fine-grained control.
WIP: Tactile & Wrist-Mounted Sensing
Integrating pressure sensors and wrist cameras to enhance manipulation understanding and fine-grained control.
Data Flow with DXTR
Data Flow with DXTR
A robot agnostic pipeline that adapts & scales with your requirements
A robot agnostic pipeline that adapts & scales with your requirements
HITL Tools for QA
HITL Tools for QA
Manual Corrections
Quickly fix errors in policies or data collection pipelines with intuitive human-in-the-loop controls.
Annotation & Trajectory Definition
Easily label objects or define motion paths to improve data accuracy and downstream performance.
Dataset Visualization & Embodiment Validation
Explore datasets visually and catch embodiment mismatches or errors before they propagate.
Manual Corrections
Quickly fix errors in policies or data collection pipelines with intuitive human-in-the-loop controls.
Annotation & Trajectory Definition
Easily label objects or define motion paths to improve data accuracy and downstream performance.
Dataset Visualization & Embodiment Validation
Explore datasets visually and catch embodiment mismatches or errors before they propagate.
Requirements to Training, in
Days not Weeks
Requirements to Training, in
Days not Weeks
A mobile app to scale your data collection efforts, across multiple modalities
A mobile app to scale your data collection efforts, across multiple modalities
Ready to train assets in standardized data formats
Ready to train assets in standardized data formats
Ready to train assets in standardized data formats
WIP
A human cloud network service, just for robotics companies - Orchestrate data procurement for any skill, at any scale.
WIP
A human cloud network service, just for robotics companies - Orchestrate data procurement for any skill, at any scale.
WIP
A human cloud network service, just for robotics companies - Orchestrate data procurement for any skill, at any scale.
Scene Segmentation & Object Annotation
Precisely label scenes and objects to power perception, planning, and manipulation workflows.
Scene Segmentation & Object Annotation
Precisely label scenes and objects to power perception, planning, and manipulation workflows.
Mobility-Aware Floor Analysis
Automatically detect obstacles, navigable regions, and floor geometry for safe lower-body movement.
Mobility-Aware Floor Analysis
Automatically detect obstacles, navigable regions, and floor geometry for safe lower-body movement.
Object Extraction & Grasp Detection
Identify and localize graspable objects with minimal human input, ready for downstream manipulation.
Object Extraction & Grasp Detection
Identify and localize graspable objects with minimal human input, ready for downstream manipulation.
Customizable Platform for Task-Specific Needs
Customizable Platform for Task-Specific Needs
Scene Segmentation & Object Annotation
Precisely label scenes and objects to power perception, planning, and manipulation workflows.
Mobility-Aware Floor Analysis
Automatically detect obstacles, navigable regions, and floor geometry for safe lower-body movement.
Object Extraction & Grasp Detection
Identify and localize graspable objects with minimal human input, ready for downstream manipulation.
Be First to Try Dxtr.
Be First to Try Dxtr.
We’re working with thoughtful design partners that are passionate about solving this problem. Together, we have a real stab at solving internet scale data collection for robots!
We’re working with thoughtful design partners that are passionate about solving this problem. Together, we have a real stab at solving internet scale data collection for robots!
© 2025 DXTR AI INC
© 2025 DXTR AI INC