Research On Human Intention Inference Based On Service Robots With Natural Human Robot Cooperation

In recent years,the demand of service robot is increasing and mainly used for interactive assembly,accompany,entertainment,companionship,helping the disabled etc.The service scene of the service robot is complicated and uncertain and there are many processes that that interact with people.In order to ensure that the service robot can complete the interaction task safely and effectively with the human in the complex daily living environment.It is necessary to improve the human-computer interaction level in order to understand the user operating intention.Most service robots can only perform instruction operations,lacking the ability to understand user intentions.The identification of user's operation intention was studied for service robot in this paper.The main research work was as follows:Therefore,it is particularly important to improve the intention inference ability of service robots and reduce the consumption of interaction resources.First of all,the appropriate RGB-D camera to get the range image of human body motion was selected and the depth information of human skeleton was obtained.The visualization of RGB-D image was realized through the depth image and RGB image coordinate transformation.On the basis of human skeletal data,the rotation angle of human joint was calculated and the motion mapping model of human body was established by analyzing the human motion map,which was based on the joint angle time series of human skeleton joint.The analytic hierarchy process(AHP)was studied and the main factors to express the user's intention was analyzed.The weight matching of the three main factors of the gestureG,the vector angle V and the distance to the target be captured S was put forward by using AHP.And the algorithm of intention recognition was proposed based on analytic hierarchy process.The user's operation intention can be predicted accurately and effectively due to its simple algorithm and low computational complexity.Finally,the intention recognition system of human-computer interaction was established and the experiments were carried out to verify the proposed algorithm of intention recognition in this paper.and two groups of experiments were designed for the purpose of grasping the target of 3 and 4 objects with different shapes.The data were collected by many experiments.Experiments showed that the user's intention can be predicted effectively by the intention recognition algorithm.Experiment of grasping the multi-target to be grasped showed that the algorithm adapted to the operation of different users and the algorithm is effective and universal.The intention recognition algorithm based on analytic hierarchy process was proposed on the basis of RGB-D image acquisition and 3D modeling of human body and the experiments was designed.Experiments showed that the user's intention can be predicted effectively by the intention recognition algorithm,showing the effectiveness and versatility of the algorithm.