Vehicle detection is essential for driverless systems. However, the current single sensor detection mode is no longer sufficient in complex and changing traffic environments. Therefore, this paper combines camera and light detection and ranging (LiDAR) to build a vehicle-detection framework that has the characteristics of multi adaptability, high real-time capacity, and robustness. First, a multi-adaptive high-precision depth-completion method was proposed to convert the 2D LiDAR sparse depth map into a dense depth map, so that the two sensors are aligned with each other at the data level. Then, the You Only Look Once Version 3 (YOLOv3) real-time object detection model was used to detect the color image and the dense depth map. Finally, a decision-level fusion method based on bounding box fusion and improved Dempster–Shafer (D–S) evidence theory was proposed to merge the two results of the previous step and obtain the final vehicle position and distance information, which not only improves the detection accuracy but also improves the robustness of the whole framework. We evaluated our method using the KITTI dataset and the Waymo Open Dataset, and the results show the effectiveness of the proposed depth completion method and multi-sensor fusion strategy.
Automated pavement distress detection benefits road maintenance and operation by providing the condition and location of various distress rapidly. Existing work generally relies on manual labor or specific algorithms trained by dedicated datasets, which hinders the efficiency and applicable scenarios of methods. Street view map provides interactive panoramas of a large scale of urban roadway network, and is updated in a recurrent manner by the provider. This paper proposed a deep learning method based on a pre-trained neural network architecture to identify and locate different distress in real-time. About 20,000 street view images were collected and labeled as the training dataset using the Baidu e-map. Eight types of distress are notated using Yolov3 deep learning architecture. The scale-invariant feature transform (SIFT) descriptors combined with GPS and bounding boxes were applied to judge the deterioration of the distress. A decision tree was designed to evaluate the change of the distress over some time. A typical road in Shanghai was selected to verify the effectiveness of the proposed model. The images of the road from 2015 to 2017 were collected from the street view map. The results showed that the mean average precision of the proposed algorithm is 88.37%, demonstrating the vast potential of applying this method to detect pavement distress. 43 distress were newly generated, and 49 previous distress were patched in the two years. The proposed method can assist the authorities to schedule the maintenance activities more effectively. INDEX TERMS Pavement distress detection, street view map, deterioration mode, deep learning, scale-invariant feature transform.
The autonomous navigation and environment exploration of mobile robots are carried out on the premise of the ability of environment sensing. Simultaneous localisation and mapping (SLAM) is the key algorithm in perceiving and mapping an environment in real time. FastSLAM has played an increasingly significant role in the SLAM problem. In order to enhance the performance of FastSLAM, a novel framework called IFastSLAM is proposed, based on particle swarm optimisation (PSO). In this framework, an adaptive resampling strategy is proposed that uses the genetic algorithm to increase the diversity of particles, and the principles of fractional differential theory and chaotic optimisation are combined into the algorithm to improve the conventional PSO approach. We observe that the fractional differential approach speeds up the iteration of the algorithm and chaotic optimisation prevents premature convergence. A new idea of a virtual particle is put forward as the global optimisation target for the improved PSO scheme. This approach is more accurate in terms of determining the optimisation target based on the geometric position of the particle, compared to an approach based on the maximum weight value of the particle. The proposed IFastSLAM method is compared with conventional FastSLAM, PSO-FastSLAM, and an adaptive generic FastSLAM algorithm (AGA-FastSLAM). The superiority of IFastSLAM is verified by simulations, experiments with a real-world dataset, and field experiments.
The battery state of charge (SOC) and state of power (SOP) are two essential parameters in the battery management system. For power lithium-ion batteries, temperature variation and the hysteresis effect are two of the main negative contributions to the accuracy of model-based SOC and SOP estimation. Thereby, a reliable circuit model is established herein to accurately estimate the working state of batteries. Considering the effect that temperature and hysteresis have on the electrical system, a unique fully-coupled temperature–hysteresis model is proposed to describe the interrelationship among capacity, hysteresis voltage, and temperature comprehensively. The key parameters of the proposed model are identified by experiments operated on lithium-ion batteries under varying ambient temperatures. Then we build a multi-state joint estimator to calculate the SOC and SOP on the basis of the temperature–hysteresis model. The effectiveness of the advanced model is verified by experiments at different temperatures. Moreover, the proposed joint estimator is verified by the improved dynamic stress test. The experimental results indicate that the proposed estimator making use of the temperature–hysteresis model can estimate SOC and SOP accurately and robustly. Our results also prove invaluable in terms of the construction of a flexible battery management system for applications in the actual industrial field.
Lithium-ion batteries (LIBs) are widely used in electric vehicles (EVs) due to their superior power performance over other batteries. However, when connected in series, overcharged cells of LIBs face the risk of explosion, and undercharged cells decrease the life cycle of the battery. Eventually, the inconsistency phenomenon between cells resulting from manufacturing tolerance and usage process reduces the overall charging capacity of the battery and increases the risk of explosion after long-time use. Research has focused on synthesizing active material to achieve higher energy density and extended life cycle for LIBs while neglecting a comparative analysis of equalization technology on the performance of battery packs. In this paper, a nondissipative equalization structure is proposed to reconcile the inconsistency of series-connected LIB cells. In this structure, a circuit uses high-level equalization units to enable direct energy transfer between any two individual cells, and dual interleaved inductors in each equalization unit increase the equalization speed of a single cell in one equalization cycle by a factor of two. The circuit is compared with the classical inductor equalization circuit (CIEC), dual interleaved equalization circuit (DIEC), and parallel architecture equalization circuit (PAEC) in the states of standing, charging, and discharging, respectively, to validate the advantages of the proposed scheme. Considering the diversity of imbalance states, the state of charge (SOC) and terminal voltage are both chosen as the equalization criterion. The second-order RC model of the LIB and the adaptive unscented Kalman filter (AUKF) algorithm are employed for SOC estimation. For effective equalization, the adaptive fuzzy neural network (AFNN) is utilized to further reduce energy consumption and equalization time. The experiment results show that the AFNN algorithm reduces the total equalization time by approximately 37.4% and improves equalization efficiency by about 4.89% in contrast with the conventional mean-difference algorithm. Particularly, the experiment results of the equalization circuit verification certify that the proposed equalization structure can greatly accelerate the equalization progress and reduce the equalization loss compared to the other three equalization circuits.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.