Shulou(Shulou.com)11/24 Report--

Thank you, Mr. Air, a netizen of CTOnews.com, for your clue delivery! CTOnews.com, March 8 (Xinhua)-- on Monday, a team of artificial intelligence researchers from Google and the Technical University of Berlin launched the largest visual language model in history, PaLM-E, with 562 billion participants (GPT-3 has 175 billion).

PaLM-E is the largest known VLM (visual language model). As a multimodal embodied VLM, it can not only understand images, but also understand and generate languages, and execute a variety of complex robot instructions without retraining. It also shows a strong ability to emerge (the model has unpredictable performance).

According to Google, when given a high-level command, such as "bring me the rice chips in the drawer," PaLM-E can generate an action plan for the mobile robot platform with arms (developed by Google Robotics) and implement these actions on its own.

PaLM-E achieves this goal by analyzing data from robot cameras without preprocessing the scene. This eliminates the need for human beings to preprocess or annotate the data, and makes the robot control more autonomous.

PaLM-E is also flexible and responsive to the environment. For example, the PaLM-E model can guide the robot to pick up a bag of potato chips from the kitchen, and because PaLM-E is integrated into the control loop, it is resistant to possible interruptions during the task. In one video example, a researcher grabs potato chips from the robot and moves them, but the robot finds the chips and grabs them again.

In addition, the PaLM-E model can also control the robot to independently complete complex tasks that require human guidance. In addition to robotics, Google researchers have observed several interesting effects of using large language models as the core of PaLM-E. One of them is that PaLM-E can show a "positive shift", which means that it can transfer knowledge and skills learned from one task to another, which performs better than a single-task robot model.

Google researchers plan to explore more real-world applications of PaLM-E in the future, such as home automation or industrial robots, and hope that PaLM-E will inspire more applications for multimodal AI.

CTOnews.com has reported that Microsoft, a fierce rival to Google's AI, recently published a paper called ChatGPT for Robotics, which combines visual data and large language models to control robots in a similar way.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.