Search

The application of robotics in gastrointestinal endoscopy has gained more and more attention over the past decade. The localization and navigation of the robotic gastrointestinal endoscopy is very important in robot-assisted gastrointestinal examination and surgery. The camera pose of the robotic gastrointestinal endoscopy can be estimated directly from the image sequence. However, due to the texture-less nature and strong specular reflections of the digestive tract surface, it is hard to detect enough keypoints to estimate the camera pose when using the traditional handcrafted method. In this paper, we propose an end-to-end CNN-based network to deal with this problem. Our network is trained in a self-supervised manner, and the network plays two roles, a dense feature descriptor and a feature detector simultaneously. The network takes the image sequence as input, and the featured keypoints and their corresponding descriptors as outputs. We demonstrate our algorithm on images captured in stomach phantom. The experimental results show that our method can effectively detect and describe the featured keypoints in challenging conditions.

Vehicular ad hoc network (VANET) has achieved promising performance in alleviating traffic problems. However, communications in VANET is vulnerable because of its open-access environment. To tackle this issue, we propose a conditional privacy-preserving authentication scheme (SACP) in this paper, which could protect communication security and driver privacy simultaneously. Specifically, symmetric encryption is used to encrypt messages, and the private keys and identity information of communication parties are employed to calculate the communication key. Moreover, only the recently exposed pseudonyms are maintained in SACP, which greatly alleviates the storage pressure of vehicles. Finally, a comprehensive empirical evaluation validates the superior performance of SACP.

Position information is one of basic elements for context awareness of autonomous mobile robots. This paper studies the positioning algorithm of autonomous mobile robots suitable for search and rescue in dark building corridors and underground mine tunnels when an emergency occurs, and proposes a novel map matching aided positioning algorithm based on a Hidden Markov Model. This algorithm does not rely on a camera, and only uses the inertial sensors installed in mobile robot and the indoor map to realize the fusion of dead reckoning and map matching. Firstly, it detects the position-related motion postures during the motion process, and then the motion trajectory is divided into a sub-trajectory sequence. By matching the sub-trajectory sequence with the indoor map, the proposed algorithm achieves tracking and positioning of the mobile robot. In order to verify the effectiveness of the proposed algorithm, this paper adopts four-wheel differentially driven robot to conduct experimental analysis in an actual indoor scenario. The experimental results show that compared with the traditional dead reckoning technology, this algorithm can distinctly reduce the average positioning error of mobile robot, and it is robust to heading angle noises within a certain error range.

Recently, Named Data Networking (NDN) has been proposed as a promising data delivery solution for smart building scenario because of its innovative properties such as name-based routing, in-network caching, and multi-source transmission. Due to the rapid growth in the number of devices and applications, it is difficult for NDN to meet the diverse transmission requirements only by leveraging the original forwarding strategy. To deal with this problem, we propose N-Guide, a novel protocol which not only supports all the transmission modes of applications but also implements the data aggregation in the scenario with large number of devices. Since our protocol directs the interest packets by the physical and application information in names just like a guide serving for the named data in smart buildings, we name it N-Guide. To verify the effectiveness of N-Guide, we make contrast simulations on ndnSIM. The results show that N-Guide outperforms two state-of-the-art NDN approaches.

In recent years, smartphone-based human activity recognition has become a promising research field of mobile computing, and is widely applied in inertial positioning, fall detection, and personalized recommendation. In practical scenario, smartphone can be placed at several body positions, such as trouser pocket, jacket pocket and so on. Since data is collected from the accelerometer embedded in smartphone, different body locations cannot generate consistent data for the same activity. As a result, the samples at a new position usually obtains low recognition rate from the classifier trained by the original data collected from other positions. In this paper, we propose a COntinuity-based POsition-adaptive recognition method, abbreviated COPO, for dealing with this problem. Considering the continuous results with high probability of correct recognition, we select them as the retraining data in COPO for updating the initial classifier. To prove the effectiveness of retraining data selecting method theoretically, we use Hidden Markov Model (HMM) to calculate the probability that the continuous recognition results are correctly recognized. Finally, a number of experiments are designed to verify our COPO, including data collection, performance comparison, and parameter analysis. The results show that the recognition rate of COPO is 2.62 % higher than other common methods.

Developing correct models for embedded systems requires formal verification. But it increases the burden on system designers to handle the verification techniques. In this paper, we solve this problem by providing a mapping from actor models to mathematical models suitable for verification; the conversion is automatic with minimal human intervention. We have integrated a model-designing Ptolemy II tool with verification tool SpaceEx by extending syntax with hybrid aspects. The integration of both tools and enhanced expressiveness allows Ptolemy II to model hybrid systems and get them verified by SpaceEx.

The smartphone-based human activity recognition method is helpful in the context awareness, health monitoring and inertial positioning. Comparing with the traditional wearable computing which fixes accelerometers on the specific positions of a user body, the activity recognition method based on a smartphone faces the problem of varying sensor locations. In this paper, we lay emphasis on the study of a feature extraction algorithm which is independent of the phone locations. First, the angle motion model is presented to illustrate the human activities. The model describes the difference among walking, going upstairs and going downstairs. Then, an angle feature extraction algorithm is proposed according to the angle motion model. Our analysis shows that different activities have significantly different angle features. Finally, our experiments are introduced. The experiments include data collecting, analysis of experiments results. The experiments results show that the recognition accuracy improved by 2% through adding the angle feature to original features.