Interpreting human motion during physical human-robot interaction is critical for enhancing robot adaptability and acceptance in collaborative tasks such as joint handling of rigid objects. Effective modeling of the trajectories of human joints from initial to target positions allows robots to adapt their motion accordingly. However, the dynamics of role shifts between human and robot during such adaptation remain insufficiently understood. This work presents a trajectory-based approach for modeling role transitions in human-robot collaboration (HRC). This method analyzes situations that trigger role-switching and interaction conflicts by evaluating force and velocity vectors at the end-effector level. A finite state machine (FSM) is implemented to classify interaction states, such as following, leading, or resisting, based on spatial deviation and motion alignment. In addition, this method monitors force unit within a desired threshold while planning robot motion along a generated human motion corridor. Unlike traditional pre-programmed robotic systems, this role-adaptive approach enables flexible and responsive behavior, supporting more natural and robust collaboration in dynamic production environments.