The recent advancements in Artificial Intelligence and robotics have highlighted the need for human-centered systems in industrial automation. As research and application of Collaborative Robots become more prevalent, it is increasingly important to ensure their ability to safely and efficiently navigate dynamic environments. This publication presents a modular ROS-based system that leverages 3D computer vision to enhance obstacle detection and avoidance. The integration of depth sensing and 3D object detection facilitates enhanced autonomy and safety in Collaborative Robot applications. The results of the tests demonstrate that the system is capable of achieving promising outcomes with low-budget sensing hardware. This provides a solid foundation for future enhancements in accuracy and flexibility, as well as the integration into other robotic systems.