← Company intelligence

Physical AI software · early

Toyota Research Institute (Robotics)

Corporate marketfoundation-modelssimulationdevtoolsCurated record

Scoped entry: covers only Toyota Research Institute's robotics AI research program (Diffusion Policy, Large Behavior Models, manipulation); parent-wide Toyota business data is not tracked.

Toyota Research Institute (TRI) is a standalone R&D organization funded by Toyota, producing foundational robotics AI: Diffusion Policy (2023), now the leading visuomotor policy learning method, outperforming prior state-of-the-art by 46.9% across 12 benchmarks. Large Behavior Models scale diffusion-policy architectures for multitask manipulation. TRI collaborates with Boston Dynamics on Atlas behavioral research, and its Punyo soft-body robot advances whole-body manipulation. TRI's outputs are licensed/shared as open research influencing commercial humanoid and manipulation companies globally.

HeadquartersLos Altos, US
Founded2015
Corporate fundingNot disclosed
TeamNot listed
Deployments0
Open roles4
01

Company overview

Identity and operating footprint

Company typePhysical AI software
Market segmentCorporate
StageCorporate
FoundersNot listed
Regions servedGlobal
Countries deployedNot listed
Service footprintNot listed
Last reviewed2026-07-05
NW

Network neighbours

Open full graph ↗
HR

Hiring signal

Public job board · 2026-07-13

Open roles4

Hiring is an operating signal, not deployment evidence. Roles are refreshed from the company’s public careers system.

TL

Company timeline

Milestones and announcements

2025-08-20Boston Dynamics and TRI demonstrate LBM-powered Atlas performing long continuous sequences of complex manipulation + locomotion tasksToyota USA Newsroom
2025-01-01TRI collaborates with Boston Dynamics on Large Behavior Models for Atlas — 450M parameter Diffusion Transformer architecture with flow-matching for whole-body manipulationBoston Dynamics
2024-10-01Toyota Research Institute and Boston Dynamics announce joint research partnership to develop general-purpose humanoids using TRI Large Behavior Models on AtlasToyota USA Newsroom
2024-01-01Diffusion Policy Policy Optimization (DPPO) published; enables zero-shot sim-to-real transfer for multi-stage manipulation tasks on physical robot hardwareToyota Research Institute
2023-09-01TRI unveils Diffusion Policy — generative AI approach to visuomotor robot control; outperforms prior SOTA by 46.9% across 12 benchmarks across 4 manipulation task setsToyota Research Institute
SC

Source ledger

5 unique public sources

Record / toyota-research-instituteLast reviewed / 2026-07-055 sourcesMachine-readable