The investigation of human body motion relies on the fundamental parameters known as Body Segment Inertia Parameters (BSIPs). Nevertheless, these quantities cannot be measured directly, hence, this paper presents a relatively convenient and swift method to estimate BSIPs of human forearm. It involves to establishing a dynamic model and identify the inertial parameters by using Recursive Least Squares (RLS) algorithm. Note that the torque generated by human muscles is difficult to be measured, thus, a platform with an active upper limb exoskeleton is designed to replace muscle effort and feedback observational data. And the purpose of trajectory planning is to mitigate abrupt variations in torque. In order to verify the feasibility of the method, three subjects are recruited for the experiment. And the estimated torque using identified inertia parameters and Kinematic information is compared to actuals. The error between the two is in the acceptable scope, which show that the proposed method can assuredly estimate BSIPs of human forearms.