Intelligent Space(IS)is widely regarded as a promising paradigm for improving quality of life through using service task processing.As the field matures,various state-of-the-art IS architectures have been proposed.Mos...Intelligent Space(IS)is widely regarded as a promising paradigm for improving quality of life through using service task processing.As the field matures,various state-of-the-art IS architectures have been proposed.Most of the IS architectures designed for service robots face the problems of fixedfunction modules and low scalability when performing service tasks.To this end,we propose a hybrid cloud service robot architecture based on a Service-Oriented Architecture(SOA).Specifically,we first use the distributed deployment of functional modules to solve the problem of high computing resource occupancy.Then,the Socket communication interface layer is designed to improve the calling efficiency of the function module.Next,the private cloud service knowledge base and the dataset for the home environment are used to improve the robustness and success rate of the robot when performing tasks.Finally,we design and deploy an interactive system based on Browser/Server(B/S)architecture,which aims to display the status of the robot in real-time as well as to expand and call the robot service.This system is integrated into the private cloud framework,which provides a feasible solution for improving the quality of life.Besides,it also fully reveals how to actively discover and provide the robot service mechanism of service tasks in the right way.The results of extensive experiments show that our cloud system provides sufficient prior knowledge that can assist the robot in completing service tasks.It is an efficient way to transmit data and reduce the computational burden on the robot.By using our cloud detection module,the robot system can save approximately 25% of the averageCPUusage and reduce the average detection time by 0.1 s compared to the locally deployed system,demonstrating the reliability and practicality of our proposed architecture.展开更多
This paper focuses on the problem of active object detection(AOD).AOD is important for service robots to complete tasks in the family environment,and leads robots to approach the target ob ject by taking appropriate m...This paper focuses on the problem of active object detection(AOD).AOD is important for service robots to complete tasks in the family environment,and leads robots to approach the target ob ject by taking appropriate moving actions.Most of the current AOD methods are based on reinforcement learning with low training efficiency and testing accuracy.Therefore,an AOD model based on a deep Q-learning network(DQN)with a novel training algorithm is proposed in this paper.The DQN model is designed to fit the Q-values of various actions,and includes state space,feature extraction,and a multilayer perceptron.In contrast to existing research,a novel training algorithm based on memory is designed for the proposed DQN model to improve training efficiency and testing accuracy.In addition,a method of generating the end state is presented to judge when to stop the AOD task during the training process.Sufficient comparison experiments and ablation studies are performed based on an AOD dataset,proving that the presented method has better performance than the comparable methods and that the proposed training algorithm is more effective than the raw training algorithm.展开更多
基金supported in part by the National Natural Science Foundation of China under Grant 62273203,Grant U1813215in part by the Special Fund for the Taishan Scholars Program of Shandong Province(ts201511005).
文摘Intelligent Space(IS)is widely regarded as a promising paradigm for improving quality of life through using service task processing.As the field matures,various state-of-the-art IS architectures have been proposed.Most of the IS architectures designed for service robots face the problems of fixedfunction modules and low scalability when performing service tasks.To this end,we propose a hybrid cloud service robot architecture based on a Service-Oriented Architecture(SOA).Specifically,we first use the distributed deployment of functional modules to solve the problem of high computing resource occupancy.Then,the Socket communication interface layer is designed to improve the calling efficiency of the function module.Next,the private cloud service knowledge base and the dataset for the home environment are used to improve the robustness and success rate of the robot when performing tasks.Finally,we design and deploy an interactive system based on Browser/Server(B/S)architecture,which aims to display the status of the robot in real-time as well as to expand and call the robot service.This system is integrated into the private cloud framework,which provides a feasible solution for improving the quality of life.Besides,it also fully reveals how to actively discover and provide the robot service mechanism of service tasks in the right way.The results of extensive experiments show that our cloud system provides sufficient prior knowledge that can assist the robot in completing service tasks.It is an efficient way to transmit data and reduce the computational burden on the robot.By using our cloud detection module,the robot system can save approximately 25% of the averageCPUusage and reduce the average detection time by 0.1 s compared to the locally deployed system,demonstrating the reliability and practicality of our proposed architecture.
基金supported by the National Natural Science Foundation of China(Nos.U1813215 and 62273203)。
文摘This paper focuses on the problem of active object detection(AOD).AOD is important for service robots to complete tasks in the family environment,and leads robots to approach the target ob ject by taking appropriate moving actions.Most of the current AOD methods are based on reinforcement learning with low training efficiency and testing accuracy.Therefore,an AOD model based on a deep Q-learning network(DQN)with a novel training algorithm is proposed in this paper.The DQN model is designed to fit the Q-values of various actions,and includes state space,feature extraction,and a multilayer perceptron.In contrast to existing research,a novel training algorithm based on memory is designed for the proposed DQN model to improve training efficiency and testing accuracy.In addition,a method of generating the end state is presented to judge when to stop the AOD task during the training process.Sufficient comparison experiments and ablation studies are performed based on an AOD dataset,proving that the presented method has better performance than the comparable methods and that the proposed training algorithm is more effective than the raw training algorithm.