Efficient Hand Gesture Recognition for Human-Robot Interaction

In this paper, we present an efficient and reliable deep-learning approach that allows users to communicate with robots via hand gesture recognition. Contrary other works which use external devices such as gloves [1] or joysticks [2] tele-operate robots, the proposed uses only visual information recognize user's instructions are encoded in a set of pre-defined gestures. Particularly, method consists two modules work sequentially extract 2D landmarks hands –ie. joints positions– predict based on temporal representation them. The has been validated recent state-of-the-art dataset where it outperformed methods multiple pre-processing steps optical flow semantic segmentation. Our achieves accuracy 87.5% runs at 10 frames per second. Finally, conducted real-life experiments our IVO robot validate framework during interaction process.