Research On Key Technologies Of Human-robot Interaction Of Commercial Service Robots

In recent years,with the breakthrough of robotics technology and the gradual maturity of application scenarios,the demand for commercial service robots is increasing.Great human-computer interaction performance is critical to the success of commercial service robots in the market.For the purpose of improving the rationality and experience of human-robot interaction,this paper studies the key technologies of human-robot interaction by investigating and analyzing the interaction problems encountered by robots in actual use.The specific research contents include: screen positioning method for service robot,the optimization of speech interaction for service robot,and design of back-end interaction scheme.The paper is organized as follows.Firstly,with respect to the fact that fixed robotic screen can't adapt to different people's heights,a screen positioning method based on the face detection algorithm is proposed.In the approach,the extracted features by HOG-ULBP is chosen as the characters to describe face information through the comparative analysis of experiments.Then,besides,considering the low accuracy of face detection in complex environment,we come up with an improved face detection algorithm,which first extracts the region of the image using GBVS-SD algorithm to reduce the influence of irrelevant regions,then obtains the HOG-ULBP features of the extracted region,and finally obtains the classification results using SVM.The experimental results show that the accuracy of the improved face detection algorithm is significantly improved in complex environments,and it can achieve screen positioning with 93.3% reliability.Secondly,some problems in the speech interaction of service robots are optimized.To enhance interactive intelligence,human body detection results are used as the condition of self-opening and self-closing of voice interaction.We implement noncontact continuous speech recognition through improved double-threshold speech endpoint detection algorithm.Moreover,by studying common speech denoising algorithms and comparing their denoising effect and time,a fast RLS algorithm is chosen as an interactive speech denoising algorithm to improve speech recognition.Thirdly,a back-end interactive server based on the Spring MVC framework is built to solve the limitation of the single platform interaction of the service robot,which realizes the long-distance command interaction and monitors the data to ensure the normal operation of the robot.Finally,the system platform is built to realize the specific functions of the Android interactive end of the robot,and various interactive functions are tested and verified on the robot entity.The test results show that the human-robot interaction mode proposed and optimized in this paper can better provide services for users.