In the past few years, we have covered plenty of robots that can navigate uneven terrain and climb stairs by using a whole host of advanced sensors and cameras. This video from Dynamic Robotics Laboratory shows how Cassie, a bipedal robot, can be taught with reinforcement learning how to climb stairs without a LiDAR or camera.
By training on hundreds of thousands of interactions with stairs in simulation, the robot can overcome all kinds of stairs and similar terrains. It can handle large steps and curbs too.
The post Blind Robot Learns to Climb Stairs with Reinforcement Learning appeared first on Robotic Gizmos.