Panasonic Holdings is set to step up research and development in “physical AI” that combines AI and robots, according to a report.
Japan’s ITmedia reported on Nov. 13 local time that Panasonic Holdings will pursue development with a team of about 250 people. It set a goal of linking the work to business results in 3 to 5 years in areas where it has strengths, such as manufacturing and logistics.
The AI and Robotics R&D Division, launched on Oct. 1, will lead the development. The company plans to develop technology with 3D spatial recognition, precise robot control and multimodal AI that processes multiple types of information together as core areas. It is also considering first verifying results at group worksites such as factories, then providing them externally as products and services.
The plan focuses on manufacturing and supply chain sites rather than immediately expanding general-purpose services. Akira Sakakibara (사카키바라 아키라), the group’s chief AI officer, cited “monozukuri and supply chain sites” as areas to focus on for the time being. He also said the company is keeping an eye on expansion into warehousing and logistics.
In manufacturing sites, the company envisions building a structure in which AI learns the experience of skilled workers. Panasonic Holdings aims to reflect in AI knowledge that is difficult to express through drawings alone, such as positional relationships between parts, friction and force adjustment needed in assembly processes. These factors are key to raising the level of automation, but are seen as areas that are not easy to standardise using existing methods alone.
The company is seeking to go a step beyond the existing process in which people review sensor-collected data and decide on improvements. It plans to develop a system in which AI directly interprets data and leads to process improvements. This is seen as an attempt to expand an approach that had stopped at visualising on-site data to the stage of process optimisation.
It is also notable that research results will be tested first on internal production lines. The company has in mind a structure in which it verifies the results at group sites such as factories and then supplies products or services to external customers. This can be seen as a strategy to narrow the gap between R&D and business application.
The organisational reshuffle is also significant in that it presupposes mid- to long-term commercialisation rather than a simple technology experiment. Panasonic Holdings said it will link outcomes to results in 3 to 5 years, and it established the AI and Robotics R&D Division separately to build a 추진 structure. In this 흐름, whether it can combine experience-based knowledge accumulated at manufacturing sites with AI and robot control is expected to be a point to watch going forward.
The company is expected to first confirm results in manufacturing and supply chains and then widen the scope of application. In particular, warehousing and logistics cited by Sakakibara are areas where robot control, spatial recognition and multimodal AI directly intersect, making them a possible first expansion area where Panasonic Holdings’ physical AI strategy could translate into an actual business.