This Robot Taught Itself How to Walk Using AI, Reinforcement Learning

researchers taught this robot to learn to walk feat.

Since the concept of robotics emerged, the long-term dream has always been humanoid robots that can live among us without posing a threat to society. Over the years, after many advancements, we have seen robotics companies create high-end robots designed for various purposes. Now, we have a pair of robotic legs that has been taught to walk on its own.

Researchers developed at the University of California, Berkley, Cassie is essentially a pair of robotic legs without a torso. At first glance, it looks a bit creepy, but when you see it learning to walk by trial and error, it looks like (the movements at least) a newborn trying to walk for the first time.

Why is Robot Cassie special?

Now, you might be thinking that we’ve already seen robots walk in the world today. Robots like Boston Dynamics’ Spot and Atlas have gained a lot of popularity on the internet, thanks to their viral video that was released earlier this year. So walking robots are old news! So what’s so special about Cassie?

Well, it turns out that choreographing a synchronized sequence of movements in robots is much easier than teaching a robot to walk alone. In the Boston Dynamics robot dance video, we have seen the robots perform in a confined space within an advanced laboratory. So, as you can imagine, it took a lot of tuning on the part of the robotics experts to program those dance moves into the robots.

However, imagine if they had to teach robots to learn to dance on their own. It would have been a difficult egg to break, right?

How did the researchers crack the code?

In Cassie’s case, the researchers used reinforcement learning to teach the machine to learn to walk on its own. It is a trial and error technique that researchers use to train the complex behavior of an AI. So, using the technique, Cassie learned a variety of moves such as crouching and walking with an unexpected load from the ground up.

The researchers used two levels of virtual environments to train Cassie. At first, they used a large database of robot movements to train a simulated version of Cassie to learn to walk on her own. They then transferred this simulation to a second virtual environment. The second virtual environment, called SimMechanics, essentially reflects real-world physics with a high degree of precision. So when the simulated version was able to walk on SimMechanics, the researchers installed the walking model on the real robot.

Once installed, Cassie was able to learn to walk on her own without any additional adjustments. During the course of the training, the pair of robotic legs were able to walk on slippery and rough surfaces, carry unexpected loads, and resist falls when pushed. During the test, Cassie resisted falling even when she damaged two engines in her right leg.

Although this all sounds exciting, Cassie is still in the early stages of development. However, robotics experts at Stanford University, Imperial College London, and Zhongyu Li, who worked on Cassie with her team, believe that these are the fundamental steps in creating advanced humanoid robots that can merge seamlessly in societies. humans in the future.

Thanks For Reading 🙂

Leave a Comment