This paper proposes a methodology called Scrumban- Ex for the development of software with uncertainty, which combines the agile Scrumban methodology and the experimental method. The characteristics of the agile methodology allow periodic deliveries of valuable versions of the software with the aim of receiving feedback from customers and in the same way, it allows efficient coordination of the activities to be carried out by the members of the development team. On the other hand, the experimental method is applied in requirements wherethe use of algorithms or novel technology is required and whose results can vary unexpectely,so the experimental method is applied to understand the behavior of the algorithms and reduce the present uncertainty. The Scrumban-Ex methodology is applied in the development of an artificial vision system dedicated to the extraction of the pose of human beings in a sequence of images. The Scrumban-Ex methodology was able to successfully satisfy the client's needs and was able to limit the development of the software to the period of time establishedat the beginning of the project.