This technology is a torque measurement system that precisely estimates joint torque by modeling and compensating for mechanical eccentricity errors within the reducer in a dual-encoder system—using both input and output encoders—and calculating the torsion angle from the angular deviation that occurs under load.
Existing robot joint torque measurement methods have faced structural limitations, such as the high cost of dedicated force/torque sensors, reduced joint stiffness in strain-gauge-based systems, and the low accuracy and difficulty of modeling reducer friction in current-based methods.
This technology proposes a method that models and stores mechanical eccentricity errors under no-load conditions using functions such as polynomials or Fourier series. During operation, it subtracts the compensation function value from the angular deviation to derive the pure torsion angle, which is then multiplied by the joint stiffness to calculate torque. It serves as an innovative solution for collaborative robots and precision assembly equipment, enabling precise force control without the need for expensive torque sensors.
This invention was developed with support from the Ministry of Trade, Industry and Energy for the development of deep reinforcement learning-based collaborative technology capable of intelligently responding to unstructured work environments, such as assembly tasks.
N/A
