Research On Gesture Interaction And Autonomous Navigation Of The Intelligent Wheelchair In Unknown Environments

With the increasing degree of social aging,the demand for intelligent products that help the elderly in their daily lives increases.As a kind of service robot for the elderly,an intelligent wheelchair has characteristics of strong environment adaptability,stable movement control and friendly man-machine interaction.Intelligent wheelchair is an inevitable development direction of intelligent products,and it has important research value and social significance.On the basis of analyzing and summarizing the research status of the intelligent wheelchair at home and abroad,an intelligent wheelchair called MIYABI ? based on natural gesture navigation in unknown environments was created.As the core of hardware platform,Kinect V2 and ARM were built.MIYABI ? prototype was developed to make up for the lack of specific target points in the current intelligent wheelchair interaction and the lack of autonomous navigation in indoor unknown environments.The innovation was to simulate human daily behavior habits,use the concept that the pointing gesture can indicate the direction.The hand was simplified,the model was built,the application was adapted in scenarios.MIYABI ? on the basic of natural gesture navigation in unknown environments was initially created.The natural gesture recognition module based on Support Vector Machine was developed,and the experiment results showed that the recognition accuracy was 99.8%.Considering the real life scenarios,some relevant kinds of natural gestures were defined command gestures and random gestures,the first 4 steps of Hu invariant moment were used as feature vectors.The gesture estimation module based on Deep Learning was studied.The average pixel error was 2.3px of the Convolutional Neutral Network model,and the forecast time was 10.0 ms for each frame.Based on the depth image information of the local hand,the 5 key points of the gesture were predicted.The point vector of the MIYABI ? user was calculated,and the target location was determined.The fuzzy obstacle avoidance and autonomous navigation modules were developed in unknown environments.The communication between human computer interaction system and motion control system was established to realize the fuzzy obstacle avoidance and autonomous navigation tasks in unknown environments.Relevant intelligent wheelchair experiments were conducted,including walking experiments driven by natural gestures,fuzzy obstacle avoidance experiments and walking experiments in unknown experiments.The experiment results showed that the feasibility and practicability of the natural gesture navigation wheelchair in unknown environments.