Research On Human-Machine Collaborative Model And Key Algorithms For Massage Robots

As the population ages,people are experiencing increasing life pressures,and elderly individuals are unable to receive adequate care.Thus,there is an increasing demand for in-home intelligent massage robots for healthcare.However,the development of such systems faces several crucial challenges to realize in-home facilities.These challenges include a lack of intelligent and autonomous massage positioning methods,fixed and single interaction modes,inconsistent behavior expressions in elderly individuals,low intent comprehension,and little initiative or decision-making capability in robots.These limitations limit the applicability of intelligent service robots in practical scenarios.Thus,with the support of the Jinan Independent Innovation Project,this study investigates human-robot collaborative massage positioning,the ability to understand intention,and safe collaboration based on natural interactions for massage robots in an applied context.This research addresses the shortcomings of current massage robots,which include limited modes of interaction,low ability to understand intention,and a lack of security.The main contributions of this article are summarized as follows:(1)This article proposes and implements a self-learning algorithm that facilitates human-robot collaborative massage positioning.The algorithm overcomes the limitations present in existing massage positioning systems by using multiple modalities to achieve accurate and flexible positioning.The candidate massage area was identified using speech,pointing,and posture modalities.The massage center point and area were then determined via roulette selection based on positioning.Self-learning was facilitated through recursive Bayesian data updating.The experimental results show that the proposed algorithm has improved positioning accuracy compared to traditional massage positioning modes.The algorithm’s natural interaction mode and self-learning capability facilitate a more fluent interaction,decrease the user’s operational burden,and enhance the user experience.(2)This article proposes and implements a multimodal fusion-based algorithm for understanding intentions.The algorithm aims to enable natural interaction and overcome the challenges of processing atypical input by analyzing implicit information from posture,speech,and gestures.Firstly,the algorithm extracts the implicit objective information from multi-modal datasets comprising gestures,speech,and postures.Subsequently,it leverages this information to perform modal data correction,multimodal fusion,and intention inference.After experimentation,the algorithm was found to reduce data uncertainty,enhance the accuracy of understanding intentions,and provide a robust framework for human-robot collaboration in massage.(3)This article presents a human-machine collaborative strategy prioritizing safety and based on comfort inference.The strategy follows a reverse proactive approach to enhance interaction safety and counteract the risks of blindly executing human commands.By firstly applying fuzzy mathematics to detect changes in user comfort after executing an intention,the strategy determines the intention’s safety.Secondly,Bayesian analysis is used to identify any intended actions that do not meet safety standards.These intentions are then either strengthened or avoided through active interaction.The strategy provides robots with decision-making agency,reduces blindly following human commands,and augments safety in human-robot interactions.In conclusion,the study developed a prototype massage robot system using a seven-axis Xarm robotic arm and a human-machine collaborative algorithm and evaluated the algorithm’s performance on the system.The experimental results demonstrate that the system can independently detect and determine appropriate massage techniques,which reduces the interaction burden for individuals seeking massages.