Many robotics applications require contact with the environment, from traditional pick and place task to legged locomotion. Nevertheless, to increase adaptability to different terrains it is necessary to know its contact properties. These properties can be known beforehand or extracted from contact forces. In this paper a practical evaluation using adaptive filtering techniques to extract the properties from different environmental conditions, with the intention to increase the adaptability of the robot to different environments is presented. We use the well-known linear spring-dashpot model and fit its parameters to four different materials in order to establish which of the reviewed methods (the recursive and windowed least squares algorithms) perform better to describe the material properties. The results show that the recursive least squares provide better tracking performance while the windowed least squares gives a smoother response.