<p>Although researchers continue to develop new methods to enhance cobot-based processes, their industrial adoption remains limited by high equipment costs and stringent environmental requirements. To address these challenges, this paper proposes an algorithm for manipulator programming under uncertain conditions. The method incorporates part position prediction in the production system using the Cullen-Frey graph, enabling adaptive modification of the cobot program in response to detected deviations. The effectiveness of the method has been demonstrated through computer-based simulations, which confirmed its robustness and scalability under varying process conditions. Experimental results further show that the application of the proposed method reduces operating time. A key advantage of this solution is that its benefits increase proportionally with the complexity of the process under consideration. The algorithm was also validated using real-world production data, confirming its capability to improve process efficiency in dynamic industrial environments.</p>