Determining the workspace of a robotic manipulator is highly significant for knowing its abilities and planning the robot application. Several techniques exist for robot workspace determination. However, these methods are usually affected by computational redundancy, like in the Monte Carlo based method case, and their implementation can be complex. The workspace analysis of kinematic redundant manipulators is even more complex. This paper proposes a kinematically optimized ray-based workspace determination algorithm based on a simple idea and not affected by computational redundancy. The proposed method can be applied to any serial robot but is tested only on spatial kinematic redundant robots. The results show how the approach can correctly determine the robot workspace boundaries in a short time. Then, the correctness and computational time of the proposed optimized ray-based method are compared to pseudo-inverse Jacobian ray-based and Monte Carlo methods. The comparison demonstrates that the proposed method has better results in a shorter time. Finally, some limitations of the proposed algorithm are discussed.