Mobile Robot Of Hand-gesture Control Based On Maps

The rapid development of internet technique and artificial intelligent technique have been pushing the update and creation in robotics. The robots are not just available in the industrial fields. They are going into our lives, being light, agile and smart. Having benefited from the progress in artificial-intelligence, service-oriented robots are blooming these days, and products with new functions are widely popular. Recently, the research on multifunctional intelligent housekeeping robots has been a hot point in this area. The multifunctional intelligent housekeeping robot is designed with the functions of domestic navigation, anomaly detection, data collection and analysis and man-machine interaction. Navigation including steps of mapping, localization, path planning, obstacle avoidance and motion control enables the robot executing its task of domestic patrol with the odometry and RGBD sensor. The two main interactive modes are speech recognition and hand gesture recognition.The work focused on in this paper mainly comprises two aspects. On the one hand, we analyze and compare the existing algorithms of hand gesture recognition, including the algorithms based on neural network, HMM and geometric features, and based on the method of using geometric features, integrating the PCANet model with contours and convex hulls, we could optimize the algorithm and improve its efficiency. Then, it will be applied to the robot for stable motion control. Another thing is to analysize and improve the SLAM algorithm, whose essence is the construction of a map, and it is a vital part of navigation. The researchers have designed some tested theories and algorithms, which have been applied in some systems. Yet, they are far from being perfect. In this paper, we would like to do some work to improve the SLAM algorithm based on Rao-Blackwellized particle filter, targeting at particle choosing and resampling steps. Finally, we design to control the robot to move domestically with a built map and localize itself with MCL, integrating the techniques of hand gesture recognition and navigation.