What is a mobile manipulation platform?
Mobile manipulators combine navigation and arm control into long-horizon episodes — the most demanding collection format in robotics. A single 30-minute demo integrates dozens of sub-tasks across a real environment.
Use cases we support:
- Household tasks — multi-room pick-and-place, loading appliances, surface cleaning
- Retail restocking — shelf manipulation, inventory handling in live environments
- Hospital logistics — cart transport, supply delivery, contactless interaction
- Office delivery — door navigation, handover, multi-floor routing
Why teams partner with us:
- 5 mobile manipulation platforms supported
- Specialist 30-minute episode operators — stamina and consistency trained
- 5,000+ trajectories collected across real-world environments
- Episode QA at the sub-task level, not just end-state verification
Long episodes demand operator stamina and consistent quality.
Our team trains specifically for 30-minute uninterrupted demos — the operator fatigue that degrades data quality in long episodes is a collection discipline problem, not just a platform problem.
30 min avg episode length
5K+ trajectories collected
5 platforms supported
Models we support
Mobile manipulators across the field
Research workhorses, quadrupeds, and production deployments.
Stretch RE-3
Hello Robot
Fetch
Research
PR2
Legacy
Spot
Boston Dynamics
Aloha Mobile
Bimanual + mobile
Tiago
PAL Robotics
What we capture
What we capture on mobile platforms
Data that captures the chains between navigation and manipulation — the part that breaks policies.
Long-horizon trajectories
Multi-step task captures, 5–30 minutes per episode, with goal annotations.
Nav + manip fusion
Combined movement and contact logs in a single timeline, your fusion stack ready.
Environment maps
Per-episode occupancy and semantic maps for retraining or replay.
Failure recovery
Operator recovery from mid-task failure, labeled for imitation or RL.
How we integrate
From platform bring-up to long-horizon set
A pipeline designed for real environments, not lab benchtops.
Platform bring-up
Robot calibration plus driver bring-up in our pod or your environment.
SLAM verification
Localization and map quality verified before production capture starts.
Long-horizon collection
Multi-step task captures with operator decision points logged.
Failure capture
Targeted re-capture of failure modes from your production model logs.
What our partners say
Explore more platforms
Six hardware families. One data partner.
Humanoid
Whole-body trajectories for bipedal robots.
Bimanual arms
Dexterous two-arm manipulation tasks.
Aerial and AMR
Navigation, inspection, and fleet data.
Related data services
Teleoperation
Remote operator-driven data collection.
Multimodal sensor capture
RGB-D, LiDAR, force, and tactile streams.
Long-tail capture
Rare scenarios your policy faces in production.
FAQ
Questions about mobile manipulation data
Scope a mobile program
Tell us the platform, the environment, and the task length. Four weeks to first long-horizon batch.