Abstract—This paper presents a preliminary robotic solution for constrained teleoperation tasks in an uncertain and dynamic environment. The robotic system is supported by a reasoning agent which makes the control action reactive and context-sensitive. The investigation is motivated by the future Human-Robot collaboration, therefore, it focuses on minimizing or avoiding collisions within the robot and the surroundings objects. The report describes the developed control architecture, which, in its modular and hierarchical structure, combines knowledge from different areas such as control theory, path and trajectory planning, computer vision, collision avoidance, and decision-making theory. The software is implemented in a ROS framework, in order to support a clear and modular design, suitable for future extensions and integration on different hardware components. The experiments are run on both real and simulated systems. The results show an autonomous robot capable of continuously adapting its movements despite the external agent interruptions, with a 99% success rate. We can conclude that an adaptive robotic system capable of performing constrained tasks and simultaneously reacting to external stimuli in an uncertain and dynamic environment is potentially obtainable.