How can AI make robots more autonomous and adaptable?

In the field of industrial automation technology, there are two recent hot spots that are difficult to ignore: artificial intelligence (AI) and Nvidia.

Don’t change the meaning of the original content, fine-tune the content, rewrite the content, don’t continue: “Not only that, the two are closely related, because NVIDIA is not limited to its most open The original graphics processing unit (GPU) is extending its GPU technology into the field of digital twins while closely connecting with emerging AI technologies. ”

Recently, NVIDIA has reached cooperation with many industrial companies, including leading industrial automation companies such as Aveva, Rockwell Automation, Siemens and Schneider Electric, as well as Teradyne Robotics and its MiR and Universal Robots Robotics Inc. Recently, Nvidia has collaborated with numerous industrial enterprises, including leading industrial automation companies such as Aveva, Rockwell Automation, Siemens, and Schneider Electric, as well as Teradyne's robotics division and its subsidiaries MiR and Universal Robots.

dealing with differences As a leader in an advanced robotics company with one of the largest installed bases of collaborative and autonomous mobile robots, Ujjwal Kumar, president of Teradyne Robotics, said the industrial sector still faces many issues. Teradyne is working with NVIDIA to help customers solve these problems. He illustrates this point with the example of autonomous pallet trucks.

There are many types of pallets used in industry. They had paint and stickers on them, and in some places had scuffed or broken wood. However, testing of automated pallet trucks is typically performed on new, nearly perfect pallets, which does not reflect the reality on most factory floors. Kumar said the industry has largely accepted this and opted to use humans to handle pallets that automated pallet trucks cannot handle.

“But what we don’t want to launch is just another autonomous pallet truck.” Kumar said, “We want to provide customers with a fully autonomous solution. But to do this, the robot needs advanced cognitive capabilities Capabilities – That’s why we partnered with NVIDIA to deliver a high level of pallet detection and security based on how it detects, responds and moves.

Kumar explained that before Nvidia’s AI capabilities were introduced into pallet inspection applications, the industry’s autonomous pallet truck capabilities were “asterisked”, meaning that they were only autonomous if the robot’s working conditions were perfect. . But now, in our new system we can say that it is an autonomous system suitable for the real world. We understand that pallets come from all over the world and they can be broken, scratched and have many defects. But our robots will no longer be looking for the perfect scenario. They will work in imperfect scenarios and poorly structured environments, with more variability than typical robotic solutions can handle.

Picture: The Jetson Edge AI module is used with the UR5e collaborative robot and vision system for quality inspection.

Software Stack

NVIDIA addresses the industry’s automation challenges through three types of computing approaches. The first is the Edge AI computer, which runs on Nvidia's Jetson platform and comes with a full software stack, explained Gerard Andrews, Nvidia's director of robotics product marketing. And then there's the big cloud computing AI - our training computers. The third is our simulation computer, which simulates the robot's performance in detail before deploying it. There is a saying that a robot spends a thousand lives in simulation before entering the real world, and we believe this is what builds trust in the robotic solution when it is finally deployed.

Nvidia also recently announced the Isaac Manipulator software stack for robotic arms and the Isaac Perceptor software stack for robot 3D vision. Andrews added that Nvidia has launched a project group focused on multi-modal AI models for humanoid robots to receive input information and generate robot actions.

Nvidia develops complete software stacks for these different types of applications, Andrews said, "because we want to understand where the limitations are. This allows our customers to get as much of the stack as they need. That's At the core of why we build things like the AI-based robotics stack and the AI-based mobile robotics stack is the reality that you don't want your solution to work only when everything is perfect. You want it to be able to handle what you're doing. Reasonable changes seen in real-world environments ”

Inflection point in advanced robotics and AI

NVIDIA’s simulation platform is changing the speed of robotics application development, Teradyne investors often ask Kumar, when will the inflection point arrive in the fields of advanced robotics and AI; that is, when will the hype around AI translate into real-world capabilities that industries need?

這個拐點可能會比在製造業20年中看到的所有技術採用速度都快。原因在於，過去任何新技術都需要在工廠中分出一部分——風險最小的部分——來嘗試新技術，同時小心不要破壞任何東西。現在，我們的大多數客戶在試行這些AI演算法時，都在雲端中的數位孿生中進行，以便在數百萬種不同的場景中進行各種測試，這將極大地加速技術的採用。因此，在Kumar看來，這個轉折點會比工業界習慣的速度快得多。

他舉出了美國北卡羅來納州和密蘇裡州的一些小型製造商與泰瑞達和英偉達合作開發應用的最新例子，這些應用現已在多個國家使用。 「在過去，只有大公司才能如此迅速地擴大規模。現在，這種擴展正是我們正在實現的。」

AI技術的融合，使機器人能夠在非結構化和不斷變化的環境中更精確地運行，並在各種規模的公司中迅速推廣新的機器人應用。

#

The above is the detailed content of How can AI make robots more autonomous and adaptable?. For more information, please follow other related articles on the PHP Chinese website!