Frank Tanner holds the role of Vice President of Robotics and Computer Vision at Innodata. Company announcements also style the position as VP of Robotics and Physical AI, as Pulse 2.0 reports in its profile interview.
Physical AI means AI systems that perceive and act in the physical world, while Innodata supplies data used to train and evaluate those systems. Tanner leads work where generative AI, data-centric systems, and real-world deployment meet. His group focuses on training data and deployment support for robotics and vision, not robots sold as standalone products.
Table of Contents
- What does Tanner lead day to day?
- Why does his background matter for automation?
- What is the New Jersey motion-capture lab for?
- How should teams cite and contact the role?
What does Tanner lead day to day?
Tanner leads Innodata's work at the intersection of generative AI, physical AI, data-centric systems, and real-world deployment. The emphasis is practical: data and evaluation methods that help AI training move into deployed operation. He directs Innodata's new motion-capture lab for robotics data.
The lab generates precise 3D body, hand, and object-interaction data by combining motion capture with synchronized video and wearable sensors. For readers, that means his remit spans both strategy and measurement. He sets the robotics-vision data direction and oversees the facility that produces ground-truth movement data.
Why does his background matter for automation?
Tanner built and led AI programs over a 30-year career across government, industry, and research. Pulse 2.0 lists mission-critical roles at AWS, Sandia National Laboratories, and Raytheon Technologies. His technical record includes work on Hubble and James Webb Space Telescope systems.
It also includes more than 20 publications, 11 AI and computer-vision patents, and service on the Applied Imagery Pattern Recognition Workshop Executive Committee. That history matters because robotics data requires calibration, validation, and deployment discipline. Teams dealing with noisy sensors, occlusion, grasp variation, or safety review gain a leader familiar with high-stakes imaging and fielded AI.
What is the New Jersey motion-capture lab for?
Innodata Investor Relations announced on Sept. 30, 2026 that the New Jersey R&D lab was developed with Vicon, as described in its lab announcement. It uses infrared optical cameras with sub-millimeter accuracy to train humanoids and industrial robots.
Tanner serves as the quoted company spokesperson for the facility. He states that direct 3D capture avoids errors from estimating depth from 2D pixels and independently validates robot-generated performance data. The practical users are teams working on sim-to-real transfer:.
- humanoid developers testing walking, reaching, grasping, and human-object interaction
- industrial robotics teams needing accurate movement references for training and evaluation
- perception teams checking vision-derived poses against independently captured 3D ground truth
How should teams cite and contact the role?
Titles vary by source, so confirm the exact remit before citing it in a proposal, article, or vendor review. Innodata describes related physical-AI data work in its company overview. Use "Vice President of Robotics and Computer Vision" unless a specific announcement uses the Physical AI wording.
For engagement, robotics teams seeking physical-AI training or evaluation data can contact Tanner's Innodata robotics and vision group. Prepare a short data request with robot type, task, environment, sensor views, and desired 3D accuracy. Include whether independent validation is needed for safety or customer acceptance. Direct 3D reference data is most valuable when simulated motion must transfer safely to real machines.



