RESUMO
Cyclists are considered to be vulnerable road users (VRUs) and need protection from potential collisions with cars and other vehicles induced by unsafe driving, dangerous road conditions, or weak cycling infrastructure. Integrating mmWave radars into cycling safety measures presents an efficient solution to this problem given their compact size, low power consumption, and low cost compared to other sensors. This paper introduces an mmWave radar-based bike safety system designed to offer real-time alerts to cyclists. The system consists of a low-power radar sensor affixed to the bicycle, connected to a micro-controller, and delivering a preliminary classification of detected obstacles. An efficient two-level clustering based on the accumulation of radar point clouds from multiple frames with a temporal projection from previous frames into the current frame is proposed. The clustering is followed by a coarse classification algorithm in which we use relevant features extracted from the resulting clusters. An annotated RadBike dataset composed of radar point cloud data synchronized with RGB camera images is developed to evaluate our system. The two-level clustering outperforms the DBSCAN algorithm, achieving a v-measure score of 0.91, compared to 0.88 with classical DBSCAN. Different classifiers, including decision trees, random forests, support vector machines (SVMs), and AdaBoost, have been assessed, with an overall accuracy of 87% for the three main object classes: four-wheeled, two-wheeled, and others. The system has the ability to improve rider safety on the road and substantially reduce the frequency of incidents involving cyclists.
RESUMO
In this paper, we address the problem of vehicle localization in urban environments. We rely on visual odometry, calculating the incremental motion, to track the position of the vehicle and on place recognition to correct the accumulated drift of visual odometry, whenever a location is recognized. The algorithm used as a place recognition module is SeqSLAM, addressing challenging environments and achieving quite remarkable results. Specifically, we perform the long-term navigation of a vehicle based on the fusion of visual odometry and SeqSLAM. The template library for this latter is created online using navigation information from the visual odometry module. That is, when a location is recognized, the corresponding information is used as an observation of the filter. The fusion is done using the EKF and the UKF, the well-known nonlinear state estimation methods, to assess the superior alternative. The algorithm is evaluated using the KITTI dataset and the results show the reduction of the navigation errors by loop-closure detection. The overall position error of visual odometery with SeqSLAM is 0.22% of the trajectory, which is much smaller than the navigation errors of visual odometery alone 0.45%. In addition, despite the superiority of the UKF in a variety of estimation problems, our results indicate that the UKF performs as efficiently as the EKF at the expense of an additional computational overhead. This leads to the conclusion that the EKF is a better choice for fusing visual odometry and SeqSlam in a long-term navigation context.
