A Topstar Technology presenter explains how TOPSTAR collects and uses real-world and simulated data to train industrial robots for autonomous work.
TOPSTAR’s in-house gripper combines vision, inertial measurements, and high-frequency sensing to track movement in six degrees of freedom. The company reports an average tracking error of 0.25%.
People demonstrate tasks such as grasping, moving items, and assembly with the gripper. TOPSTAR’s digital platform aligns and standardizes the resulting data, then groups it into training sets.
The datasets train strategies that connect what a robot sees with how its end effector should move. TOPSTAR says the model is refined continuously, then linked to a control system that uses live camera input to carry out tasks independently.
TOPSTAR records robot operations and examines cases where performance goes wrong. Useful new examples return to collection and training, creating a feedback cycle; real-world capture and simulation jointly expand the data supply.
Briefs summarise talks and interviews; timestamps open the video at that moment. Claims in a talk are the speaker's own.
Topstar Technology’s TopStar discusses with Loon AI an industrial physical AI data platform for robot handling and how it can support robot training, evaluation, and deployment.
