Souce
Climb a volcano more than 6,000 m high and face one of the most hostile and freezing environments on the planet. Forget lab videos where robots just dance or execute multiple perfectly choreographed challenges, robotics and mountaineering engineering has reached an impressive milestone.
On terrain with inclinations of less than 30º, the robot walked alone and maintained good stability. However, during the most technical and steep sections of the long 16-hour journey to the summit, the team members themselves had to carry the machine on their backs. The goal now is to use reinforcement learning to train the neural networks directly with the data collected on this expedition, allowing the metal legs to learn to traverse ice, rocks and cracks completely independently.
Posted Using INLEO
