TECHNOLOGY AND INNOVATION CENTER

Seki: The main difference is that THINKLET® is a neck-mounted device rather than smart glasses or a head-mounted display (HMD) device. Using devices that are worn on the head in a real work environment can be cumbersome. A neck-mounted device is more comfortable and convenient for on-site use, allowing for hands-free operation. Additionally, head-mounted devices tend to cause image instability due to frequent head movement, but the neck-mounted design minimizes this issue, providing stable and less distracting visuals.

 ――How might THINKLET® change the world? Besides real-time remote support, what other applications could it have?

Seki: One potential application is utilizing recorded videos. It's important to consider that we live in a world in which information can be categorized into easily text-encoded (verbal) information and difficult-to-text-encode non-verbal information. Recently, it has become easier to extract and organize information from text using large language models (LLMs). However, much of the information and skills that are not text-encoded still exist on-site, and we need to capture that as well. For instance, video footage of hands-on tasks performed on-site and the conversations that occur during those tasks hold significant value.

The ultimate goal is to use AI to analyze data captured with THINKLET® and transform beginners into multi-skilled workers.

Although there are still various challenges, we believe that having workers wear THINKLET® all day to accumulate data, detect important scenes, and build knowledge is a promising approach. The value lies in how we can utilize non-verbal information hidden in the video and use it as an interface. However, accumulating video data requires time and effort, and it's crucial to convey the significance of this accumulation to on-site workers.

As a first step, we plan to start with human-assisted remote support, then move on to partial task automation, and eventually provide insights through search and analysis of recorded videos. Daikin is already using remote support extensively and has accumulated thousands of hours of data. Our dream for the future is to achieve "multi-skilled workers" where beginners can perform any task using THINKLET® based on the accumulated data.

Application Examples Only Possible with Daikin: Its First-Person Video Analysis Leads Competitors

Over the past year, Daikin has been accumulating first-person work videos while simultaneously providing remote work support. First-person video analysis enables task procedure assistance, safety support, and replication of optimal procedures.