Decreasing user effort and automating subtasks such as obstacle avoidance and user guidance has shown to increase the effectiveness and utility of teleoperation. Extending the capabilities of teleoperation remains a critical research topic for tasks that need to leverage user knowledge, or for unstructured environments that autonomous solutions are not robust enough to handle. Previous methods have focused individually on joint space tasks, regression or training based user intention recognition and intervention, or application specific solutions. To overcome the limitations of these methods, this paper proposes the use of path planning based gross motion assistance with a projection based user intention recognition method, for improving task execution in semi-autonomous teleoperation. The proposed solution synthesizes an assistive architecture that leverages the benefit of supervisory level task identification with semi-autonomous trajectory tracking. With the proposed method, continuous and more immersive teleoperation is achieved, as control states are user selected and task execution is informed from the operator’s motion. The effectiveness of the proposed method is validated with a user study.