Toyota’s $6.4bn robotics estimate puts physical AI in focus

Toyota Motor projects that rolling out automation across its manufacturing facilities, subsidiaries, and key suppliers may require approximately 400,000 robots and an annual expenditure of roughly 1 trillion yen ($6.4 billion) starting in 2028.

According to Reuters, this potential investment was presented to investors in early September. Toyota has not confirmed whether the full investment will move forward or how long spending at this scale would be sustained.

The 400,000 robot estimate accounts for both replacements of existing machinery and new installations, encompassing humanoid and non-humanoid systems. This investment covers industrial robots, automated logistics, and future collaborative robotics on factory floors.

Toyota currently utilizes and tests robotics in various manufacturing stages. In its FY2026 financial results, the company highlighted factory demonstrations involving parts transportation and picking, as well as robotics applications in medical equipment logistics.

Building on previous automation initiatives

At Toyota’s Kamigo Plant, a piston assembly line that launched with robotic assistance in January 2025 has transitioned from three operators to full automation, as noted in Toyota’s 2025 integrated report.

The plant’s automation journey began much earlier. It started deploying robots on selected sub-lines in 2008, but the company later faced maintenance challenges as engine production expanded overseas.

Toyota noted that some overseas facilities lacked sufficient skilled maintenance staff to support automated equipment. Consequently, some processes reverted to manual production while manufacturing expertise from Japanese plants was transferred to overseas sites.

Workers then developed jigs and tools that Toyota stated could double or triple efficiency in certain manual processes. The knowledge gained from this work informed the development of newer automated systems, including the piston assembly line at Kamigo.

Toyota is also creating systems capable of handling parts without requiring precise pre-positioning. Its KumiPro parts-picking robot uses cameras to identify and manipulate loosely positioned components and is already active on production lines at Toyota Motor East Japan.

Toyota has demonstrated an assembly process where force-feedback control compensates for camera recognition errors. The robot adjusts its movement while inserting components into tight spaces, accounting for discrepancies between visual recognition and physical positioning.

Developing physical AI for manufacturing

Toyota defines physical AI as technology enabling robots to perceive their environment via sensors, learn or optimize actions based on situational data, and operate autonomously.

Toyota stated that its Frontier Research Center addresses labor shortages and skill transfer through robotics, including production site applications. The center’s work includes technologies derived from its Human Support Robot program.

Toyota indicated that object recognition, decision-making, and motion-execution technologies from the Human Support Robot program have been integrated into KumiPro at Toyota Motor East Japan.

Toyota is also developing ELEY (Embodied Learning robot for Enhanced Yield) for tasks involving physical interaction with objects and the environment. Equipped with two human-like arms and an omnidirectional mobile base, ELEY is designed to handle external forces when a machine encounters an object at an unexpected position.

Toyota has identified long-duration reliability, precision repeatability, and the data infrastructure for robot learning as areas requiring further development. The company plans to test ELEY under conditions closer to actual production sites, using both successful and failed attempts as training data.

Separately, Toyota uses reinforcement learning to train humanoid robots in simulated environments before transferring learned behaviors to physical machines. The company runs thousands of simulated robot instances in parallel to accelerate the learning process.

Toyota noted that movements successful in simulation do not always transfer directly to physical hardware due to differences in sensor readings, floor friction, and actuator behavior. Researchers introduce variations in these conditions during training and use data from physical robots to reduce what Toyota calls the Sim2Real gap.

Toyota Research Institute is also collaborating with Boston Dynamics on AI control systems for the Atlas humanoid robot. In 2025, the organizations demonstrated Atlas completing a sequence of walking, lifting, sorting, and packing tasks using a single Large Behavior Model, with robot skills acquired through human demonstrations.

Japan was the world’s second-largest market for industrial robots in 2024, with approximately 450,500 robots in operation and another 44,500 installed during the year, according to the International Federation of Robotics.

Japan’s automotive industry installed about 13,000 industrial robots in 2024, an 11% increase from the previous year and the highest annual total since 2020.

Toyota’s 400,000 estimate is broader than IFR’s industrial robot figures because it includes replacement equipment and multiple robot types across Toyota, its group companies, and major suppliers.

Hyundai Motor Group has established a timeline for deploying humanoid robots in automotive production. It plans to introduce Atlas at Hyundai Motor Group Metaplant America in Savannah, Georgia, starting in 2028, initially for parts sequencing before expanding into component assembly from 2030.

Hyundai stated it will expand Atlas into additional manufacturing processes after validating individual applications. The group has also set a target to establish annual production capacity for up to 30,000 robots by 2028.

Boston Dynamics, the developer of Atlas, is controlled by Hyundai Motor Group, which acquired a controlling stake in the robotics company in 2021. Boston Dynamics also collaborates separately with Toyota Research Institute, combining Atlas with TRI’s Large Behavior Models.

See also: Gartner outlines four AI tiers in warehouse automation

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and is co-located with other leading technology events including the Cyber Security & Cloud Expo. Click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.