Eureka, artificial intelligence for training robots

A artificial intelligence model Developed by an American technology company, Nvidia teach robot Until now, these complex abilities have been within the realm of human capabilities.

This artificial intelligence agent is powered by GPT-4 and write independently reward algorithm Train any type of robot such as quadrupeds, bipeds, quadcopters, dexterous hands and collaborative robotic arms (Cobots); in activities such as opening drawers, using scissors, catching balls and nearly 30 different tasks.

This innovative reinforcement learning is a machine learning This allows the agent to learn from its own actions and feedback.The best example of these features is how eureka I can teach one hand robot Perform tricks with pens just like people.

Researchers claim that the reward algorithms generated by Eureka are much better than those written by human experts 80% tasks, resulting in average performance improvements exceeding Fifty% For robots.

This artificial intelligence works alongside simulation technology to accelerate graphics processor from NVIDIA, such as Isaac Gym, a physics simulation reference application for reinforcement learning research; and Omniversea development platform for building tools and 3D applications framework based open dollar.

Tool that allows this AI model to quickly assess the quality of large batches of award candidates More effective training.

In fact, Eureka can build a Summary of key statistics Based on the training results and instructing its language model to improve the generation of its reward function.In this way, artificial intelligence self-improvement.

Researchers say Eureka was developed new algorithm Integrating generative and reinforcement learning methods to solve difficult tasks. They also predict that the technology will be able to control dexterous robots and provide artists with a new way to create physically realistic animations.

The idea of ​​robots teaching robots is generating growing interest and success.an article published in a magazine Transactions on Machine Learning Research A new system was launched called Skills (shared knowledge for lifelong learning)which enables AI systems to learn 102 different skills

including diagnosing disease and identifying flower species through chest X-rays.

These models share knowledge as teachers through communication networks and are able to master each of the 102 skills.In addition, researchers from schools, e.g. MIT and University of Bristol They’ve also had success, particularly in using artificial intelligence to teach robots to manipulate objects.

However, the combination of artificial intelligence and robotics has raised various fears and concerns in society, such as:

– Unemployment and worker replacement: One of the most prominent concerns is that AI-driven automation could lead to the loss of jobs for human workers. Robots and automated systems have the potential to perform repetitive and routine tasks more efficiently than humans, which could lead to certain jobs becoming obsolete.

As artificial intelligence advances, people worry that it will surpass not only human physical strength, but also human cognition and creativity. This could impact professionals in fields such as medicine, law and artistic creativity.

– Control and autonomy: There are legitimate concerns that AI systems and robots may become too autonomous and escape human control. Scenarios where autonomous robots make decisions on their own, especially in military applications, may raise concerns about a lack of control and oversight.

– Security and cyberattacks: Artificial intelligence in robotics is vulnerable to cyberattacks and hacking. If autonomous or semi-autonomous systems fall into the wrong hands, they could be used for malicious purposes, raising national security and privacy concerns.

– Unethical development: Concerns about the unethical development of artificial intelligence are a major concern. If not properly implemented and regulated, AI can be used inappropriately or discriminatoryly, which can have a negative impact on society.

-Over-reliance: Overreliance on AI and automation may lead to a decrease in people’s ability to perform manual or cognitive tasks themselves, which may negatively impact human autonomy and capabilities.

– Prejudice and discrimination: AI systems can learn biases and stereotypes present in training data. This can lead to discriminatory or unfair decisions in applications such as personnel selection or judicial decision-making.