The localization of coal mine robots (CMRs) serves as the foundation for intelligent mines. Despite various approaches proposed by academia and industry, existing range-based localization methods generally encounter the non-line-ofsight (NLOS) problem, leading to severe accuracy degradation in complicated underground mines. Facing the above challenge, this paper proposes an edge-assisted cross-domain NLOS localization (CrossDNL) framework based on semi-supervised learning. Specifically, we analyze the channel impulse responses (CIRs) of ultra-wideband (UWB) to identify the multi-channel NLOS conditions and further mitigate ranging errors via the deep neural network (DNN). To achieve reliable localization across various scenarios, CrossDNL adopts an edge-assisted semi-supervised learning architecture that enables it to optimize DNN models by self-training on the edge server and achieve the real-time service on the CMR. Besides, we propose a range-based error state Kalman filter (ESKF) scheme to improve the localization performance. We implement CrossDNL on the CMR and the edge server and evaluate it in various scenarios. Results show that CrossDNL can achieve a 14cm localization error in real time, outperforming the latest solutions by more than 20%.

Edge computing provides a promising solution to the resource-constrained problem of mobile devices running simultaneous localization and mapping (SLAM) systems. However, the existing traditional edge-assisted SLAM systems rely on ideal network conditions and static environment assumptions, which poses challenges for applications in bandwidth constrained networks and dynamic scenes. Aiming at the issue, this paper proposes fast and accurate edge-assisted semantic SLAM (FAES-SLAM) towards dynamic environments. First, an edge-assisted framework is presented by the integration of object detection and dense mapping, which accelerates the SLAM system and maintains accuracy. Then, an adaptive keyframe offloading strategy and an efficient dynamic feature point culling scheme are designed and incorporated into the above framework to improve the speed and accuracy further. The offloading strategy adaptively decreases the excessive latency in terms of various network conditions, and the culling scheme effectively increases the localization and mapping precision for dynamic scenarios. We perform extensive experiments on public datasets and in the actual coal mine laboratory. Compared with classical edge-assisted SLAM system, FAES-SLAM can achieve over 90.9% improvement in localization accuracy with a 13.2% increase in tracking time.

Aiming at problems of under-fitting and poor model robustness in learning-based wireless sensing methods caused by the lack of large-scale wireless sensing datasets, this paper proposes a privacy-friendly collaborative wireless sensing framework, called Co-Sense. It builds a community with multiple clients and a server, which aggregates the clients' local models into a federated model with cross-domain capability. To protect the privacy of users' local data, we innovatively introduce the idea of federated learning into the field of wireless sensing, by uploading users' local model parameters instead of their local data. Then, in response to the uneven computing power of different users' edge devices, we propose a local model update algorithm based on adaptive computing power. Furthermore, a client selection algorithm based on test nodes is designed to reduce the negative influence of malicious clients on Co-Sense. Finally, we evaluate Co-Sense on three well-known public wireless datasets, including the gesture dataset, the activity dataset, and the gait dataset. Experimental results show that the sensing accuracy of Co-Sense is more than 10% higher than that of the most advanced wireless sensing models.

The prevalence of illumination equipment and the inherent advantages of the Visible Light Communication (VLC) technique have resulted in a growing interest in Visible Light Positioning (VLP). There exist many excellent VLP techniques over the past several years. However, one limitation of most VLP survey works is that they mainly focus on the analysis from the perspective of techniques but ignore the equally important hardware aspect, since the hardware part directly affects the performance and cost of VLP systems and also determines whether it can be put into practical use. Different from most surveys concentrating on a single perspective, we provide an intensive overview of VLP systems from software algorithms to hardware devices. A novel‐innovative classification method is used in the software algorithms, while the hardware aspect is introduced in terms of transmitters, modems, and receivers, making up for the deficiencies of the previous works. Massive papers including pioneering papers and the state‐of‐the‐art ones in related areas are gathered and categorized. These solutions have also been evaluated in terms of accuracy, cost, range, and complexity. Furthermore, current open issues and tendencies regarding VLP are also illustrated in this paper.

Carpooling is becoming a more and more significant traffic choice, because it can provide additional service options, ease traffic congestion, and reduce total vehicle exhaust emissions. Although some recommendation systems have proposed taxicab carpooling services recently, they cannot fully utilize and understand the known information and essence of carpooling. This study proposes a novel recommendation algorithm, which provides either a vacant or an occupied taxicab in response to a passenger’s request, called VOT. VOT recommends the closest vacant taxicab to passengers. Otherwise, VOT infers destinations of occupied taxicabs by similarity comparison and clustering algorithms and then recommends the occupied taxicab heading to a close destination to passengers. Using an efficient large data-processing framework, Spark, we greatly improve the efficiency of large data processing. This study evaluates VOT with a real-world dataset that contains 14747 taxicabs’ GPS data. Results show that the ratio of range (between forecasted and actual destinations) of less than 900 M can reach 90.29%. The total mileage to deliver all passengers is significantly reduced (47.84% on average). Specifically, the reduced total mileage of nonrush hours outperforms other systems by 35%. VOT and others have similar performances in actual detour ratio, even better in rush hours.