Building the data layer
for robot intelligence.
Radioception was founded on a simple premise: the bottleneck to capable robots is not algorithms — it's high-quality training data. We built the infrastructure to capture, annotate, and deliver that data at enterprise scale.
Human expertise, machine-readable.
When a robot needs to learn to grasp a mug, fold a shirt, or insert a bolt, it needs to see expert human hands doing exactly that — thousands of times, from multiple angles, with every joint position tracked to sub-millimeter precision.
That's what Radioception does. We're the production house for robot training data — staffed by motion capture engineers, annotation scientists, and robotics researchers who understand what your model actually needs.
We've delivered over 14,800 datasets to 340+ robotics companies across grasping, assembly, locomotion, surgical robotics, and more.
Leadership Team
Dr. Anika Reyes
CEO & Co-founder
Previously: Director of Robotics Research, DeepMind. PhD Computer Vision, MIT.
James Okonkwo
CTO & Co-founder
Previously: Principal Engineer, Scale AI. MS Robotics, Carnegie Mellon.
Dr. Lena Fischer
Head of Annotation Science
15 years motion capture research. Developed Radioception's QA validation pipeline.
Marcus Webb
Head of Capture Operations
Former motion capture supervisor, Industrial Light & Magic. 12 years MoCap expertise.
Priya Sharma
VP of Engineering
Previously: Staff Engineer, Waymo. Expert in robotics data pipelines and ROS integration.
Dr. Carlos Vega
Research Scientist
PhD Biomechanics, Stanford. Specializes in human motion modeling and annotation protocols.
Capture Technology Stack
Video Capture
- ◆Sony FX6 (4K60 UHD)
- ◆GoPro Hero 12 (body cams)
- ◆Intel RealSense D455 (depth)
- ◆Stereolabs ZED 2i (stereo depth)
Motion Capture
- ◆Vicon Vantage 16-cam rig
- ◆Xsens MTw Awinda (IMU)
- ◆Mediapipe Hand Landmarks
- ◆OpenPose full-body
Sensor Suite
- ◆ATI Nano17 force-torque
- ◆ATI Mini45 (wrist mount)
- ◆Ouster OS1 LiDAR
- ◆Realsense L515 (structured light)
Annotation Pipeline
- ◆Labelbox (video annotation)
- ◆Custom KineticAnnotate UI
- ◆Automated confidence scoring
- ◆Human-in-the-loop QA
Data Infrastructure
- ◆AWS S3 (encrypted storage)
- ◆AES-256 at rest & in transit
- ◆Private VPC delivery links
- ◆Version control for datasets
Output Formats
- ◆Egocentric Raw Video (.mp4)
- ◆Kinematics Video (.mp4)
- ◆JSON format (.json)
- ◆Parquet format (.parquet)
- ◆RLDS tfrecords (.tfrecord)
Enterprise-Grade Data Handling
SOC 2 Type II
Independently audited security controls for data handling, availability, and confidentiality.
NDA by Default
All production engagements are covered by mutual NDAs. IP exclusivity available for enterprise clients.
AES-256 Encryption
All data encrypted at rest and in transit. Secure VPC delivery links with expiring access.
GDPR Compliant
Demonstrator consent collected per GDPR guidelines. Full data subject deletion available on request.