Remote-controlled VR robots to assist Japanese supermarket workers

The young Japanese company Telexistence will soon deploy semi-autonomous and remotely controlled employee robots in several konbini, mini-markets present on every street corner in Japan. Where some countries are banking on the rise of online commerce or the development of cashless stores, such as Switzerland or the United States, Japan is counting on robotics to automate its retail sector.

In this video, we can see a person operating a Model-T five miles away.

The Model-T is a 22-DOF robot (22 different movements) remotely controlled using a virtual reality headset and equipped with a hand flexible enough to allow handling any type of merchandise. The objective of the device is not to replace humans entirely, but to assist them in restocking store shelves. According to Telexistence, the video transmission latency (for a 2160X1200 at 80 fps file encoded in H.265 / HEVC) between the robot’s camera and the operator’s screen is only 50 milliseconds. The data passes through mobile network, LAN or WiFi via the AWP platform (Augmented Workforce Platform), hosted on Microsoft Azure.

“Although robots only sell a million units per year compared to 100 million cars, we believe there is great potential in the field. As a start-up, you have to create momentum to change the direction of the company: we are thus inspired by the impact that Ford Motor has had with its economically affordable T-models, ”explains Jin Tomioka, CEO of Telexistence.

The robots will be tested at several FamilyMart stores in Tokyo, with the goal of rolling them out in at least 20 stores by 2022. Rival brand Lawson will also test the Model-T at select locations in September. While no pricing has yet been announced, it should be low according to Telexistence. According to Family Mart, a successful deployment of these robots will increase employee flexibility, since they will not have to be on site to work, and will also help alleviate the employee shortage that affects the entire sector.