In the study of intelligent space oriented to home service robot, an important technology is how to construct an communication network which has the characters of high reliability and easy building. In this paper, bas...In the study of intelligent space oriented to home service robot, an important technology is how to construct an communication network which has the characters of high reliability and easy building. In this paper, based on the characteristics of ZigBee protocol, ZigBee technology is used to construct a wireless sensor and actor network. Several intelligent services based on ZigBee wireless sensor and actor network are shown to certify the reliability of this communication network. ZigBee wireless sensor and actor network builds an information bridge for the components in the intelligent space, the spatially distributed devices are connected together seamlessly. With this network, robot can share the mass information in the intelligent space and improve its performance with 'light-packs', devices in intelligent space, such as lamp, curtain can be controlled autonomously.展开更多
In order to improve the learning ability of robots, we present a reinforcement learning approach with a knowledge base for mapping natural language instructions to executable action sequences. A simulated platform wit...In order to improve the learning ability of robots, we present a reinforcement learning approach with a knowledge base for mapping natural language instructions to executable action sequences. A simulated platform with physical engine is built as interactive environment. Based on the knowledge base, a reward function with immediate rewards and delayed rewards is designed to handle sparse reward problems. Also, a list of object states is produced by retrieving the knowledge base, as a standard to define the quality of action sequences. Experimental results demonstrate that our approach yields good performance on accuracy of action sequences production.展开更多
In order to satisfy the robotic personalized service requirements that can select exclusive items to perform inference and planning according to different service individuals,the service robots need to have the abilit...In order to satisfy the robotic personalized service requirements that can select exclusive items to perform inference and planning according to different service individuals,the service robots need to have the ability to independently obtain the ownership relationship between humans and their carrying items.In this work,we present a novel semantic learning strategy for item ownership.Firstly,a human-carrying-items detection network based on human posture estimation and object detection model is used.Then,the transferred convolutional neural network is used to extract the characteristics of the objects and the back-end classifier to recognize the object instance.At the same time,the face detection and recognition model are used to identify the service individual.Finally,on the basis of the former two,the active learning of ownership items is completed.The experimental results show that the proposed ownership semantic learning strategy can determine the ownership relationship of private goods accurately and efficiently.The solution of this problem can improve the intelligence level of robot life service.展开更多
With the development of artificial intelligence and robotics, the study on service robot has made a significant progress in recent years. Service robot is required to perceive users and environment in unstructured dom...With the development of artificial intelligence and robotics, the study on service robot has made a significant progress in recent years. Service robot is required to perceive users and environment in unstructured domestic environment. Based on the perception,service robot should be capable of understanding the situation and discover service task. So robot can assist humans for home service or health care more accurately and with initiative. Human can focus on the salient things from the mass observation information. Humans are capable of utilizing semantic knowledge to make some plans based on their understanding of the environment. Through intelligent space platform, we are trying to apply this process to service robot. A selective attention guided initiatively semantic cognition algorithm in intelligent space is proposed in this paper. It is specifically designed to provide robots with the cognition needed for performing service tasks. At first, an attention selection model is built based on saliency computing and key area. The area which is highly relevant to service task could be located and referred as focus of attention(FOA). Second, a recognition algorithm for FOA is proposed based on a neural network. Some common objects and user behavior are recognized in this step. At last, a unified semantic knowledge base and corresponding reasoning engine is proposed using recognition result. Related experiments in a real life scenario demonstrated that our approach is able to mimic the recognition process in humans, make robots understand the environment and discover service task based on its own cognition. In this way, service robots can act smarter and achieve better service efficiency in their daily work.展开更多
The indoor robots are expected to complete metric navigation tasks safely and efficiently in complex environments, which is the essential prerequisite for accomplishing other high-level operation tasks. 2 D occupancy ...The indoor robots are expected to complete metric navigation tasks safely and efficiently in complex environments, which is the essential prerequisite for accomplishing other high-level operation tasks. 2 D occupancy grid maps are sufficient to support the robots in avoiding all obstacles in the environments during navigation. However, the maps based on normal laser scans only reflect a horizontal slice of the environment, which may cause the problem of some obstacles missing or misinterpreting their exact boundaries,thereby threatening the safety and efficiency of robot navigation. This paper presents a 2 D mapping method based on virtual laser scans to provide a more comprehensive representation of obstacles for indoor robot navigation. The resulting maps can accurately represent the top-down projected contours of all obstacles no matter where their vertical positions are. The virtual laser scans are initially generated from raw data of an RGB-D camera based on the filtering, projection, and polar-coordinate scanning. The scans are fed directly to the laser-based simultaneous localization and mapping(SLAM) algorithms to update the current map and robot position. Two auxiliary strategies are proposed to further improve the quality of maps by reducing the impact of the narrow field of view and the blind zone of the RGB-D camera on the observations. In this paper, the improved virtual laser generation method makes the extracted 2 D observations fit the laser-based SLAM algorithms, and two auxiliary strategies are novel ways to improve map quality. The generated maps can reflect the comprehensive obstacle information in indoor environments with good accuracy. The comparative experiments are carried out based on four simulation scenarios and three real-world scenarios to prove the effectiveness of our 2 D mapping method.展开更多
文摘In the study of intelligent space oriented to home service robot, an important technology is how to construct an communication network which has the characters of high reliability and easy building. In this paper, based on the characteristics of ZigBee protocol, ZigBee technology is used to construct a wireless sensor and actor network. Several intelligent services based on ZigBee wireless sensor and actor network are shown to certify the reliability of this communication network. ZigBee wireless sensor and actor network builds an information bridge for the components in the intelligent space, the spatially distributed devices are connected together seamlessly. With this network, robot can share the mass information in the intelligent space and improve its performance with 'light-packs', devices in intelligent space, such as lamp, curtain can be controlled autonomously.
基金supported by National Natural Science Foundation of China (No. 61773239)Shenzhen Future Industry Special Fund (No. JCYJ20160331174814755)
文摘In order to improve the learning ability of robots, we present a reinforcement learning approach with a knowledge base for mapping natural language instructions to executable action sequences. A simulated platform with physical engine is built as interactive environment. Based on the knowledge base, a reward function with immediate rewards and delayed rewards is designed to handle sparse reward problems. Also, a list of object states is produced by retrieving the knowledge base, as a standard to define the quality of action sequences. Experimental results demonstrate that our approach yields good performance on accuracy of action sequences production.
基金This work was supported by the Joint Funds of National Natural Science Foundation of China(Nos.U1813215 and 2018YFB1307101)National Natural Science Foundation of China(Nos.61603213,61773239,61973187,61973192 and 91748115)+2 种基金Shandong Provincial Natural Science Foundation,China(No.ZR2017MF014)Jinan Technology project(No.20150219)Taishan Scholars Programme of Shandong Province.
文摘In order to satisfy the robotic personalized service requirements that can select exclusive items to perform inference and planning according to different service individuals,the service robots need to have the ability to independently obtain the ownership relationship between humans and their carrying items.In this work,we present a novel semantic learning strategy for item ownership.Firstly,a human-carrying-items detection network based on human posture estimation and object detection model is used.Then,the transferred convolutional neural network is used to extract the characteristics of the objects and the back-end classifier to recognize the object instance.At the same time,the face detection and recognition model are used to identify the service individual.Finally,on the basis of the former two,the active learning of ownership items is completed.The experimental results show that the proposed ownership semantic learning strategy can determine the ownership relationship of private goods accurately and efficiently.The solution of this problem can improve the intelligence level of robot life service.
基金supported by National Natural Science Foundation of China (Nos. 61773239, 91748115 and 61603213)Natural Science Foundation of Shandong Province (No. ZR2015FM007)Taishan Scholars Program of Shandong Province
文摘With the development of artificial intelligence and robotics, the study on service robot has made a significant progress in recent years. Service robot is required to perceive users and environment in unstructured domestic environment. Based on the perception,service robot should be capable of understanding the situation and discover service task. So robot can assist humans for home service or health care more accurately and with initiative. Human can focus on the salient things from the mass observation information. Humans are capable of utilizing semantic knowledge to make some plans based on their understanding of the environment. Through intelligent space platform, we are trying to apply this process to service robot. A selective attention guided initiatively semantic cognition algorithm in intelligent space is proposed in this paper. It is specifically designed to provide robots with the cognition needed for performing service tasks. At first, an attention selection model is built based on saliency computing and key area. The area which is highly relevant to service task could be located and referred as focus of attention(FOA). Second, a recognition algorithm for FOA is proposed based on a neural network. Some common objects and user behavior are recognized in this step. At last, a unified semantic knowledge base and corresponding reasoning engine is proposed using recognition result. Related experiments in a real life scenario demonstrated that our approach is able to mimic the recognition process in humans, make robots understand the environment and discover service task based on its own cognition. In this way, service robots can act smarter and achieve better service efficiency in their daily work.
基金supported by National Natural Science Foundation of China(Nos.U1813215 and 61773239)the Taishan Scholars Program of Shandong Province(No.ts201511005)。
文摘The indoor robots are expected to complete metric navigation tasks safely and efficiently in complex environments, which is the essential prerequisite for accomplishing other high-level operation tasks. 2 D occupancy grid maps are sufficient to support the robots in avoiding all obstacles in the environments during navigation. However, the maps based on normal laser scans only reflect a horizontal slice of the environment, which may cause the problem of some obstacles missing or misinterpreting their exact boundaries,thereby threatening the safety and efficiency of robot navigation. This paper presents a 2 D mapping method based on virtual laser scans to provide a more comprehensive representation of obstacles for indoor robot navigation. The resulting maps can accurately represent the top-down projected contours of all obstacles no matter where their vertical positions are. The virtual laser scans are initially generated from raw data of an RGB-D camera based on the filtering, projection, and polar-coordinate scanning. The scans are fed directly to the laser-based simultaneous localization and mapping(SLAM) algorithms to update the current map and robot position. Two auxiliary strategies are proposed to further improve the quality of maps by reducing the impact of the narrow field of view and the blind zone of the RGB-D camera on the observations. In this paper, the improved virtual laser generation method makes the extracted 2 D observations fit the laser-based SLAM algorithms, and two auxiliary strategies are novel ways to improve map quality. The generated maps can reflect the comprehensive obstacle information in indoor environments with good accuracy. The comparative experiments are carried out based on four simulation scenarios and three real-world scenarios to prove the effectiveness of our 2 D mapping method.