Rocky is still learning to keep its footing. At the National Research Council, researchers are training the knee-high, four-legged robot to handle slopes, rough ground and other changing conditions. The team uses reinforcement learning and simulated landscapes before moving to real-world testing. Cameras, LiDAR and foot-force sensors help the robot judge its surroundings and maintain balance. Researcher Abhishek Naik’s goal is adaptable autonomy: a robot that can respond when circumstances change. Disaster recovery, mining, agriculture and lunar exploration are possible future applications. The October 28 report describes ongoing research, not a robot already working independently in those environments.
