Building and programming your own robots is a great way to learn skills. You don’t need fancy parts to build one either. Take the MEVITA for instance: it is an open source bipedal robot assembled with off-the-shelf components. Unlike many other bipedal robots of this type, this one doesn’t need to be 3D printed. Its makers relied on sheet metal welding to build this.
Thanks to reinforcement learning in simulation and Sim-to-Real transfer, this robot learned walking across various terrains. As you can see in the video, this robot can move on grass and other uneven surfaces. It is based on a Jetson Orin Nano Developer Kit with a LIVOX MID-360 module.
[HT] [credit: K. Kawaharazuka and S. Sawaguchi and A. Iwata and K. Yoneda and T. Suzuki and K. Okada]
