Physical AI Companies

Pre-event WRC 2026 material announced roughly 300 exhibitors and 2,000 exhibits across 4 zones. Use that planned scope to trace the physical AI supply chain.

EASA’s June 2026 Easy Access Rules integrate SORA 2.5 AMC and GM for specific-category UAS operations. SORA structures risk evidence; it is not an approval, and transition treatment depends on the competent authority.

DAAAM links objects, 3D space, time, events, and language in a hierarchical 4D scene graph. Its benchmark improvements do not guarantee that a household robot will always find a moved object or preserve privacy.

MIT’s Masked IRL combines demonstrations with two language-model steps to identify task-relevant state features before inverse reinforcement learning. Its reported gains are bounded experiments, not general instruction understanding.

KUKA AMP is presented as an operations layer linking AI-agent intent to robot, AMR, and digital-twin state. Its early Toledo AMR scope does not mean that AMP replaces PLCs, safety control, or every vendor interface.

UR AI Trainer records leader–follower demonstrations with synchronized motion, force, and video. Useful imitation-learning data also needs calibration, failure labels, intervention history, rights, and a separate policy evaluation gate.

O4 Ground Station combines 12 antennas, multiple regional bands, gateway or relay modes, and long-range claims. Unattended inspection still needs a full link, backhaul, power, interference, and recovery budget.

Wing and Walmart announced seven additional US metro areas and a path toward more than 270 locations in 2027. Current service, rollout plans, Wing’s one-million-delivery milestone, and local address eligibility must be kept separate.

Skydio reports 1,070 deployed Docks and more than four million company-wide flights. Those milestones establish scale, but not the proportion of fully autonomous missions, availability, incident rate, or customer-level performance.

Agras T55 and T100 should be compared as field workflows, not tank-size rankings. Payload, battery rotation, spray and spread rates, terrain sensing, refill travel, and local rules determine useful capacity.