Clustering and Regression to Impute Missing Values of Robot Performance

It is widely claimed that a major challenge in Robotics is to get reliable systems while both response and down times are minimized. In keeping with this idea, present paper proposes the application of a Hybrid Artificial Intelligence System (HAIS) to preprocess data with the aim of improving the detection of performance anomalies. One of the main problems when analyzing real-life data is the presence of missing values. It is usually solved by removing incomplete data, what causes a loss of information that may be critical in some domains. As an alternative, present paper proposes the application of regression models to impute those missing values. Prediction is optimized by generating personalized models on previously clustered data. Experiments are run on a public and up-to-date dataset that contains information about anomalies affecting the component-based software of a robot. The obtained results validate the proposed HAIS, as it successfully imputes missing values from the different features in the original dataset.