Google's DeepMind team has taken a significant leap in robotics safety by introducing a 'Robot Constitution,' inspired by Isaac Asimov's famous Three Laws of Robotics.
This innovative approach aims to ensure that AI-powered robots, designed for a variety of tasks, operate safely and effectively in human environments.
Isaac Asimov explains his three laws of robotics in this 1965 BBC Horizons programme ✍️pic.twitter.com/x1Bcif3qQo— Physics In History (@PhysInHistory) September 28, 2023
AutoRT - The Heart of Safe Robotics
At the core of this initiative is Google's AutoRT system, a groundbreaking technology that combines the prowess of visual language models (VLMs) and large language models (LLMs).
"The system uses a VLM to understand its environment and the objects within sight. Then, an LLM suggests a list of tasks, acting as the decision-maker," Google explains in its blog post.
This dual approach allows the robots to adapt to new settings and choose appropriate tasks, all while maintaining safety as the top priority.
The 'Robot Constitution' in Action
The Robot Constitution is more than a theoretical framework; it's a set of practical, safety-focused prompts guiding the LLMs.
These rules are designed to prevent robots from engaging in tasks involving humans, animals, sharp objects, or electrical appliances.
DeepMind's safety measures go beyond software algorithms.
The robots are programmed to halt automatically if the force on their joints exceeds a specific threshold.
Additionally, human supervisors are equipped with a physical deactivation switch for added security.
This is how the new homegrown robots will start with foundation from Google. I think we are ready for this. pic.twitter.com/Lp7SSUHbFU— Madhu Reddiboina (@MReddiboina) January 5, 2024
While we may still be years away from robots serving drinks or cooking meals autonomously, Google's advancements with AutoRT bring us closer to this reality.
As we move towards a future where robots are an integral part of our homes and workplaces, Google's approach, grounded in Asimov's visionary laws, offers a promising blueprint for creating machines that not only assist but coexist safely with humans.