This media is not supported in your browser
VIEW IN TELEGRAM
Robo-Dog Conquers Vertical Ladders
A method has been developed to train a quadruped robot to climb ladders. By combining hook-like grippers with reinforcement learning algorithms, researchers achieved a 90% success rate on slopes ranging from 70° to 90°. The climbing speed has been increased by 232 times compared to previous models.
The robot adapts to different ladder parameters and remains stable against external disturbances. Remarkably, the system requires no additional tuning when transitioning from simulation to a real robot. The climbing gait resembles that of humans: the hind legs generate lifting force while the front legs provide stability.
Pros: No need to walk it.
Cons: Can't hide, can't escape.
#robotics #climbingrobots
@science
A method has been developed to train a quadruped robot to climb ladders. By combining hook-like grippers with reinforcement learning algorithms, researchers achieved a 90% success rate on slopes ranging from 70° to 90°. The climbing speed has been increased by 232 times compared to previous models.
The robot adapts to different ladder parameters and remains stable against external disturbances. Remarkably, the system requires no additional tuning when transitioning from simulation to a real robot. The climbing gait resembles that of humans: the hind legs generate lifting force while the front legs provide stability.
Pros: No need to walk it.
Cons: Can't hide, can't escape.
#robotics #climbingrobots
@science
👀41👍35🔥10😁8⚡3🕊2