Andrew Ng and his team have introduced Agentic Object Detection, an innovative AI system that can identify objects without any prior training. Users can simply provide a text prompt along with an image, allowing the AI to think, plan, and locate the specified object. This system is comparable to the capabilities of OpenAI's o1/o3 models and aims to enhance object detection processes by eliminating the need for extensive training or labeling data. In a related development, Chinese researchers have created a reinforcement learning framework that enables humanoid robots to stand up in various real-world scenarios, addressing challenges faced by robots due to rigid training methods.
Chinese researchers developed a reinforcement learning framework that enables humanoid robots to stand up in diverse real-world scenarios https://t.co/Vprw4xOQ0K
Recently in #AI+#Robotics: Zero-shot approach allows robots to manipulate articulated objects https://t.co/8XeuAVzLhV
Learning Humanoid Standing-up Control across Diverse Postures. https://t.co/sXBKDSnqjR