Nouvelle AI

Nouvelle artificial intelligence, or Nouvelle AI, is a special way of creating thinking machines that started in the 1980s. It was created by Rodney Brooks, who worked at the MIT artificial intelligence laboratory. This idea is different from older types of artificial intelligence because it tries to make robots act smart like tiny insects.

People who work with Nouvelle AI think that being smart doesn’t have to come from complex plans. Instead, they believe smart actions can grow naturally when simple behaviors meet the real world. This is unlike older methods that needed everything to be planned out in special, made-up worlds. By letting robots act and react in real situations, researchers hoped to see how intelligence could develop on its own.

This approach helped change how we think about making smart machines and robots, focusing more on learning through action rather than just following strict rules.

Motivation

The differences between nouvelle AI and symbolic AI can be seen in early robots like Shakey and Freddy. These robots had an internal model of their small, made-up worlds made of symbols. This meant they had to update their symbols constantly as they moved or as the world changed.

Shakey's programs took a long time to plan and break down tasks, so it worked very slowly. Symbolic AI had trouble updating and managing these symbol worlds inside robots. Nouvelle AI is different because it looks at the real world through sensors instead of using an internal model. As Rodney Brooks said, “the world is its own best model—always exactly up to date and complete in every detail.”

A key idea in nouvelle AI is that simple actions can come together to form more complex actions over time. For example, basic actions like “move forward” and “avoid obstacles” can help a robot learn to do something more complicated, like following a moving object.

The frame problem

The frame problem talks about a challenge when using first-order logic to describe facts about a robot. Traditional methods need lots of rules to say what does not change in the environment.

Nouvelle AI tries to avoid this by not filling the robot with lots of rules. Instead, it lets more complex actions develop naturally from combining simpler ones.

Embodiment

Traditional AI aimed to create intelligence that had no body and could only interact with the world through a keyboard, screen, or printer. Nouvelle AI, however, tries to build embodied intelligence that exists in the real world. Brooks liked ideas from Alan Turing’s sketches in 1948 and 1950, where Turing suggested giving a machine the best sensors possible and teaching it to understand and speak English, similar to how a child learns. This was different from other approaches that focused on abstract tasks like playing chess.

Brooks' robots

Brooks worked on building robots that acted like simple insects. He created robots named Allen and Herbert, inspired by two important thinkers in the field of thinking and learning machines.

Allen used special sensors to stay in the middle of a room and move away from objects that came near. Herbert used different sensors to find empty soda cans in busy office areas and carry them away. These actions happened because of many simple behaviors working together.

Brooks also made other robots, like Genghis, which could walk over rough ground, and Squirt, which would follow sounds. Later, Brooks tried to build a robot named Cog that could learn like a human, but work on this stopped in 2003.