On April 27, Daniel Zhang, Alibaba‘s Chairman and CEO and CEO of Alibaba Cloud Intelligence, revealed at the 6th Digital China Summit that Alibaba Cloud engineers are experimenting with integrating the Qianwen large-scale model into industrial robots. Just by entering a sentence in human language into the DingTalk chat box, remote control of robot work can be achieved.
A demonstration video released by Alibaba Cloud at the event showed this experimental achievement. After an engineer instructed the robot through the DingTalk chat box saying “I’m thirsty, find something for me to drink,” Qianwen immediately understood the instruction and replied: “Okay, I’ll see what there is to drink.” Then the large-scale model automatically wrote a set of code and sent it to a robot. The robot began recognizing its surroundings and found a bottle of water on a nearby table before completing a series of actions such as moving, grabbing and delivering it smoothly to the engineer.
Zhang said that “manufacturing is an important battlefield for AI large-scale models. The biggest opportunity in cloud computing, AI and physical world machine integration over the next ten years lies here. Robots fetching water is just a start. Intelligent robots capable of direct dialogue with humans will change entire factory operations.”
SEE ALSO: Alibaba Cloud Announces the Largest Price Reduction in Its History
At present, the control of robots is heavily dependent on codes. Operating robots necessitates individuals to learn complex programming languages or acquire knowledge about robotic systems. Nevertheless, large-scale models like GPT offer novel possibilities for breaking down barriers between humans and AI by utilizing multi-modal language models as a means of communication between humans and robots.
In addition, large-scale models not only support applications but also participate in development stages within robotics fields.
According to an Alibaba Cloud engineer interviewed during this event, when developing robots, engineers can use Qianwen’s large-scale model-generated code instructions for developing and debugging robotic functions while even creating some new features for them.
