Human face orientation recognition for intelligent mobile robot collision avoidance in laboratory environments using feature detection and LVQ neural networks

In this paper, an approach on the intelligent mobile robot collision avoidance is proposed for the complex laboratory robot transportation process using the human face orientation recognition strategy. The proposed approach includes the contents as: (a) Measuring the face images of laboratory personnel by the adopted Microsoft Kinect sensors; (b) Processing the measured face images to recognize the face orientations which will be used to control the mobile robots in the collision avoidance; and (c) Building the Learning Vector Quantization (LVQ) Neural Networks to calculate and decide the face orientations based on the extracted face feature data. To select the best training algorithm for the LVQ model, a trail experiment is provided in the study. The results of the study show that: based on a standard laptop, the successful rate and the elapsed time of the proposed human face recognizing method are 99% and 3.17s, respectively. It means the proposed method can be applied in the mobile robot collision avoidance applications.