This study presents a new six-DOF force/torque sensor and its calibration method for a “collaborative robot or cobot”. This new calibration method applies the so-called “maximum likelihood estimation (MLE)”. MLE is used to determine and identify the parameters related to applied torques/forces and resulted deformations at varied locations of the sensor structure. Formulating the relations in a vector-matrix form, such parameters are captured as the coefficients in a matrix relating torques/forces to angular/linear deformations in different directions. In addition to applying MLE, finite element (FEM) modeling and analysis are conducted to generate realistic-like empirical data for the aforementioned calibration computation based on MLE. The matrix formed by these coefficients can then predict forces and torques with good accuracy via deformations detected by strain gauges. Seen in simulation results, the worst sum of error for three forces is less than 0.04%, while that for three torques is less than 0.00005%. Very small values of sum of error demonstrate a strong correlation of the model with the results obtained from MLE, indicating high precision in sensing forces and torques without decoupling matrix. The developed sensor is highly suitable for real-time sensing of multidimensional interactive forces/torques in industrial robots.