How does iCub support studies of sensorimotor learning?
The iCub research platform was created to study embodied learning by relating vision, touch, joint position and actions. Researchers can use its open software ecosystem to study grasping and interaction, but published results depend on the hardware generation, installed hands and exact control pipeline. A model trained from one robot's tactile readings may not reproduce on a differently calibrated hand. Cognitive robotics tasks also require measured behavior rather than labels such as understanding.