We’ve all seen what some of the four-legged robots already on the market are capable of and what awaits us when bipeds perfect their techniques, but they all consist of a series of movements that require a lot of work and practice beforehand. .
What if it was the robot itself that could make its own decisions about the terrain on which it should move? This is the idea with which an MIT team has created a robot dog capable of recognizing the environment in which it finds itself, and they used the M1 chip to achieve this.
Why be on a Mac if you can be on a dog?
What you see in the video above is the result of the MIT artificial intelligence laboratory: a four-legged robot capable of overcoming obstacles in the terrain on which it must move, using only a front camera and an algorithm capable of to process its progress. using neural volumetric memory.
If you notice, the robot has a little white box on its back. inside her, according to the authors, inhabits the motherboard of a Mac with an M1 chip whose cores are used in the necessary neural network be able to do all the data processing. No details are given about the Mac they had to to scrap to be able to do it or how they integrated the board to make it the “brain” of the robot, although promised disclose it with the source code used on Github. One thing is certain: the warranty for this Mac fell through.
The advantage is that the robot does not need prior field work to recognize the ground: it only recognizes it and decides which steps to take without falling. They have all the information along with several other GIFs on this website. And if the M1 chip is able to process all the data without problems, imagine how far we could go with an M2 Pro or one of the future M3s.
In Applesphere | This user’s iPhone 14 Pro Max is broken, broken and bent in half. Always works perfectly
