OLogic CEO and co-founder Ted Larson will headline a technical session on egocentric robot learning at RoboBusiness 2026, the industry conference set for October 20 to 21 at the Santa Clara Convention Center. His talk, "Egocentric Robot Learning: Teaching Robots Through Human Experience," argues that first-person video from wearable cameras — combined with teleoperation and imitation learning — is now the fastest way to give a robot a useful skill outside a research lab.
Why egocentric data is the moment
The last twelve months have seen NVIDIA, Google DeepMind and Meta all release large-scale physical-AI stacks that lean heavily on human demonstration data — from Cosmos world models and GR00T policies to Meta's Ego-Exo research and Aria wearables. Larson's argument is that this convergence gives smaller robotics teams a way out of the traditional data-collection bottleneck: capture a warehouse operator, hotel host or hospital courier doing the job with a head-mounted camera, and use that video as the training signal instead of paying labelers by the frame.
OLogic's client footprint sets the stage
OLogic is a robotics design and contract-engineering studio that has quietly shipped robots for a long list of brands, including Bear Robotics, Locus Robotics, Simbe Robotics, Misty Robotics, Zoetic, Hasbro and Panasonic. That vantage point matters for the session: Larson can compare what teleoperation and egocentric learning actually cost across payloads and duty cycles that range from a home toy up to Simbe's Tally shelf-scanner and Bear's hospitality service robots.
F-Prime brings the money view
F-Prime Capital partner Sanjay Aggarwal and principal Betsy Mule will present the venture firm's annual State of Robotics Investments report on the same day, giving the conference a rare pairing of an operator perspective on how robots are being trained with a top-of-funnel view of where robotics capital is actually flowing. F-Prime's 2026 report has become a widely cited benchmark for physical AI investment; a preview is expected to land ahead of the show.
Deployment math is the real story
Egocentric learning is not academic for OLogic's clients. Labor shortages in warehouses, restaurants and hospitals have compressed the time customers are willing to wait between purchase order and productive robot — often to weeks, not quarters. Larson frames egocentric data collection as the piece that lets a general-purpose behavior model be specialized on the customer's floor without months of custom engineering.
Related coverage on The Robotics Media: Agility Robotics Digit 5 unveil, InOrbit.AI OpenRobOps ISO 21423 launch, and Dyna Robotics scaling law paper.
Reporting based on coverage from The Robot Report and OLogic.