Digital Event Horizon
A New Era of Physical Intelligence: How AI is About to Revolutionize the Real World
In 2025, we can expect a seismic shift in the way machines interact with and adapt to their surroundings. According to experts, this new era of physical intelligence will enable devices, from robots to power grids and smart homes, to not only understand digital instructions but also execute tasks in the real world.
Researchers at Carnegie Mellon University have developed a robot that can perform dynamic and complex parkour movements using just one camera and imprecise actuation. AI systems currently have limitations in their ability to navigate and interact with the physical world, which is rooted in their digital nature. The development of physically intelligent machines requires a fundamental shift in how machines think and interact with the environment. Liquid networks are being developed as a new form of intelligence that can interpret and physically execute complex commands from text or images. These models have shown to be more adaptable than traditional AI systems, learning and adapting from experience like humans do. Other labs are making breakthroughs in the field of physical intelligence, such as developing chatbots that can control robotic arms.
In a recent breakthrough, researchers at Carnegie Mellon University have demonstrated that a robot with just one camera and imprecise actuation can perform dynamic and complex parkour movements. This achievement paves the way for the development of physically intelligent machines capable of navigating and interacting with the physical world.
According to Daniela Rus, director of the Computer Science and Artificial Intelligence lab at MIT, the current limitations of AI systems are rooted in their digital nature. "Artificial intelligence models remain surprisingly humanlike in their ability to generate text, audio, and video when prompted," she noted. "However, so far these algorithms have largely remained relegated to the digital world, rather than the physical, three-dimensional world we live in."
The development of physically intelligent machines requires a fundamental shift in how machines think and interact with the environment. This new form of intelligence is rooted in physics and understanding the fundamental principles of the real world, such as cause-and-effect.
In her research group at MIT, Rus is developing models of physical intelligence known as liquid networks. These models are capable of interpreting and physically executing complex commands derived from text or images, bridging the gap between digital instructions and real-world execution.
For instance, in one experiment, two drones were trained to locate objects in a forest during the summer using data captured by human pilots. While both drones performed equally well when tasked with exactly what they had been trained to do, only the liquid network drone successfully completed its task when asked to locate objects in different circumstances – during the winter or in an urban setting.
This experiment demonstrated that traditional AI systems stop evolving after their initial training phase, whereas liquid networks continue to learn and adapt from experience, much like humans do. This property of liquid networks has significant implications for the development of physically intelligent machines capable of interacting with a wide range of environments and tasks.
Other labs are also making groundbreaking breakthroughs in the field of physical intelligence. Robotics startup Covariant, founded by UC-Berkeley researcher Pieter Abbeel, is developing chatbots that can control robotic arms when prompted. They have already secured over $222 million to develop and deploy sorting robots in warehouses globally.
Furthermore, a team at Carnegie Mellon University has demonstrated that a robot with just one camera and imprecise actuation can perform dynamic and complex parkour movements – including jumping onto obstacles twice its height and across gaps twice its length. This achievement highlights the potential of physically intelligent machines to navigate and interact with their surroundings in a wide range of contexts.
In conclusion, the development of physically intelligent machines is set to revolutionize the way we live and work. With AI gaining physical intelligence in 2025, we can expect a new era of devices that not only understand digital instructions but also execute tasks in the real world. As researchers continue to push the boundaries of what is possible with these machines, we can look forward to a future where technology seamlessly integrates with our daily lives.
Related Information:
https://www.wired.com/story/ai-physical-intelligence-machine-learning/
Published: Tue Jan 7 20:42:53 2025 by llama3.2 3B Q4_K_M
