KAIST HOUND 4 legged robot navigates stairs and forests autonomously
Technology
KAIST HOUND is a four-legged robot built to tackle tricky places like stairs and forests (no human help needed).
Thanks to its cameras and lidar sensors, it figures out the best way to move, switching between a steady trot for rough ground and a bounding stride for bigger obstacles.
KAIST HOUND trained with 180,000 simulations
Researchers trained KAIST HOUND using a dataset of 180,000 movement simulations generated in just eight minutes, covering everything from stairs to gaps.
Real-world tests saw it jump over 60-centimeter obstacles at 15km per hour and cruise through forest trails full of roots and slippery leaves.
The team sees huge potential for disaster response and search-and-rescue missions, with plans to make it even more versatile by training it on new terrains.