When the robot is plugged in, the edge is repeatedly rubbed but it is difficult to fit in place. When holding a plastic bottle, the bottle body is depressed due to an imbalance of force — in industrial scenarios, it is not uncommon for intelligent robots to “roll over” in industrial scenarios. The crux is that robots must not only rely on visual perception to interact with the physical world; multi-modal information such as touch also plays an important role. Taking robots from “seeing” the world to “touching” the world, tactile perception is nurturing a rapidly growing segmented circuit. According to incomplete statistics, the sector added more than 7 billion yuan in financing in the first seven months of this year, and at least two companies became new unicorns. The Securities Times reporter learned in an interview that incorporating tactile information into large models of embodied intelligence to improve the flexible operation ability of robots has become one of the mainstream exploration directions in the industry. Relying on technological achievements such as visual touch integration, flexible operation in the robot industry has developed application test scenarios, and large-scale mass production is expected to be launched. However, due to the irreplaceable nature of tactile data, the data gap is still one of the core bottlenecks that hinder large-scale implementation of robots. The industry calls for further improvement of technical routes such as data simulation while strengthening the actual collection of multi-modal data; core components such as tactile sensors need to strive for excellence in durability and data consistency.

When the robot is plugged in, the edge is repeatedly rubbed but it is difficult to fit in place. When holding a plastic bottle, the bottle body is depressed due to an imbalance of force — in industrial scenarios, it is not uncommon for intelligent robots to “roll over” in industrial scenarios. The crux is that robots must not only rely on visual perception to interact with the physical world; multi-modal information such as touch also plays an important role. Taking robots from “seeing” the world to “touching” the world, tactile perception is nurturing a rapidly growing segmented circuit. According to incomplete statistics, the sector added more than 7 billion yuan in financing in the first seven months of this year, and at least two companies became new unicorns. The Securities Times reporter learned in an interview that incorporating tactile information into large models of embodied intelligence to improve the flexible operation ability of robots has become one of the mainstream exploration directions in the industry. Relying on technological achievements such as visual touch integration, flexible operation in the robot industry has developed application test scenarios, and large-scale mass production is expected to be launched. However, due to the irreplaceable nature of tactile data, the data gap is still one of the core bottlenecks that hinder large-scale implementation of robots. The industry calls for further improvement of technical routes such as data simulation while strengthening the actual collection of multi-modal data; core components such as tactile sensors need to strive for excellence in durability and data consistency.