THE DATA LAYER FOR PHYSICAL AI
The world is your training ground.
We collect egocentric, UMI and teleoperation
data around the tasks your robots need to learn.
Egocentric. UMI. Teleoperation.
One partner for your data pipeline.
SAMPLE DATA
01 / THE MISSING LAYER
Your next breakthrough needs more of the world: the way people move, manipulate objects, and get things done.
Telepath connects your models to that experience. We collect robot training data around your tasks, with the right people, capture methods, and environments.
How robotics data collection works Why the data matters more than the model02 / OUR DATA
01 / EGO / DATA ACQUISITION
First-person video of people getting things done. Capture the context, object interactions, and sequences behind everyday work.
For learning from human behavior, task understanding, and visual representations of the world.
02 / UMI / DATA ACQUISITION
Manipulation demonstrations captured with handheld grippers. Bring visual observations and gripper motion together around the task.
For teams building manipulation policies from demonstrations, with capture requirements aligned to the target setup.
03 / TELEOP / DATA ACQUISITION
Human-guided robot operation turns task execution into data. Collect observations and actions through the robot your team needs to study.
For embodiment-specific demonstrations, task refinement, and learning from human intervention.
03 / OUR REACH
Telepath is based in Dubai and collects in real workplaces across the United Arab Emirates and the CIS, with sites in Kazakhstan, Uzbekistan, Tajikistan. The data is built for robotics and physical-AI teams in the United States, led by the San Francisco Bay Area.
04 / THE TASK LANDSCAPE
From a single grasp to a complete workflow. Explore example manipulation tasks we can scope for your next collection.
Every project starts with your task definition. Capture setup, volume, and site availability are agreed together.
Select different objects from bins and place them into order totes. Vary object geometry, bin layouts, and grasp approaches.
Object selection · Grasping · Placement
Move parcels between containers using visual sorting criteria. Scope different packaging materials, sizes, and work surfaces.
Visual sorting · Transfer · Task sequences
Open containers, arrange items, and handle packaging materials through a complete packing or unpacking sequence.
Container interaction · Bimanual handling · Sequencing
Arrange plates, cups, and cutlery to a defined layout. Include different table settings and object placements in the brief.
Spatial arrangement · Object orientation · Placement
Manipulate towels and soft textiles through sequential folds, with variation in material, dimensions, and starting configuration.
Deformable objects · Bimanual actions · Task progression
Clear cups and objects from a work surface and return them to assigned locations, capturing the order and context of each action.
Scene understanding · Object transfer · Workflow completion
Transfer products from a tote to shelves across varied heights, layouts, and product shapes.
Reach variation · Object handling · Shelf placement
Reposition products and align them within a shelf display, with different orientations and starting arrangements.
Object orientation · Fine repositioning · Visual alignment
Group mixed products into bins according to a task brief, including varied packaging and visual selection criteria.
Categorization · Selection · Placement
Select components and assemble a task-specific kit, varying part types, tray positions, and kit requirements.
Component selection · Precise placement · Task sequences
Pick up, orient, use, and return tools within an agreed workstation task. Align the capture setup to the actions your team needs.
Tool orientation · Controlled manipulation · Work sequences
Capture how people pick up, rotate, and inspect parts at a workstation, with the inspection criteria defined in the collection brief.
Multi-view observations · Object rotation · Inspection sequences
EXAMPLE COLLECTION BRIEFS · CUSTOM TASKS WELCOME
05 / HOW WE WORK
You define what your model needs to learn. We work backwards from that goal to build the collection, in four stages.
The full processAlign on tasks, capture methods, environments, and acceptance criteria. Define what a useful demonstration looks like before collection begins.
YOUR OUTPUT / A SHARED COLLECTION BRIEF
Coordinate site access, people, and capture setups around the agreed task. Collect the variations and sequences your model needs.
YOUR OUTPUT / TASK-ALIGNED DEMONSTRATIONS
Review captures against the agreed criteria. Identify gaps in the task coverage, flag collection issues, and align on what needs another pass.
YOUR OUTPUT / REVIEWED COLLECTION DATA
Receive the collection in the dataset structure and format agreed with your team, with the task context and metadata defined in the brief.
YOUR OUTPUT / DATA STRUCTURED FOR YOUR TEAM
06 / WHO IS BEHIND THIS
Telepath Robotics is based in Dubai, United Arab Emirates, with site access across the Middle East and the CIS, and customers among robotics teams in the United States. Pilot programmes are open for teams that need task-specific robot training data.
About the companyWHAT A DELIVERY LOOKS LIKE
Every collection is delivered in the layout and format agreed in the brief. The example below shows one common structure: episodes grouped by task, each with synchronised observations, actions where applicable, and the metadata defined at the Define stage.
collection-2026-09-table-setting/
├── brief.md # agreed tasks, criteria, sites
├── manifest.json # episode index, durations, review status
├── episodes/
│ ├── ep_0001/
│ │ ├── video_ego.mp4 # first-person capture
│ │ ├── gripper.csv # UMI pose + width (when in scope)
│ │ ├── actions.parquet # teleop commands (when in scope)
│ │ └── meta.json # site, operator id, variation tags
│ └── ep_0002/ …
└── review/
└── coverage-report.md # gaps found, re-collection notesEXAMPLE LAYOUT · FORMATS FOLLOW YOUR BRIEF
07 / INSIGHTS
Essays on training data, distribution shift, perception, and the path from human demonstration to robot capability.
All insightsFROM THE WORLD. INTO YOUR MODEL.
Let’s build the data behind their next capability.
Build your dataset