Robotic roller forming is a promising sustainable manufacturing technique, but it faces significant forming errors, primarily due to the low stiffness of serial industrial robots. This study developed a robotic roller forming platform using the KUKA KR600 industrial robot equipped with a six-dimensional force/torque sensor. A stiffness-based error compensation method was proposed to improve the processing accuracy. During the forming process, posture, force, and torque data were collected and then the deviation in the end-effector was calculated using the established stiffness model of the industrial robot. Based on the estimated deviation in the roller, the operational trajectory of the end-effector was compensated during the final pass of the rolling process. Experimental results demonstrate that the proposed compensation method reduced forming errors by 58.4%, from 3.45 to 1.44 mm. This enhancement substantially improves manufacturing precision, demonstrating its potential to lower the rejection rate of products and minimize processing steps, thereby conserving resources.