Abstract: A charging method for a secondary battery including a lithium-supplementing material. The method includes acquiring a first state of health of the secondary battery in response to the secondary battery being at a preset charging node, activating the lithium-supplementing material in response to the first state of health being less than or equal to a first threshold to supplement lithium for the secondary battery, performing a charging process on the secondary battery, determining a second state of health of the secondary battery based on a working parameter of the secondary battery in the charging process, and charging the secondary battery in response to the second state of health being greater than a second threshold.

Abstract: A method and a device for self-supervised learning, a storage medium, and an electronic device are provided. The method includes: organizing real points in one column along a vertical direction into a pillar; determining a predicted point in a next frame; determining a first loss term based on a minimum distance among distances between predicted points in the next frame and real points in the next frame, and generating a loss function including the first loss term; and performing self-supervised learning processing based on the loss function. A pillar motion parameter representing motion of a real point is determined with the pillar as a unit, so as to enhance correlation between point clouds. Self-supervised learning can be realized in a case of no precise correspondence between the predicted point and the real point, and training is performed based on a large number of unlabeled point clouds.

Abstract: A charging method for a secondary battery including a lithium-supplementing material. The method includes acquiring a first state of health of the secondary battery in response to the secondary battery being at a preset charging node, activating the lithium-supplementing material in response to the first state of health being less than or equal to a first threshold to supplement lithium for the secondary battery, performing a charging process on the secondary battery, determining a second state of health of the secondary battery based on a working parameter of the secondary battery in the charging process, and charging the secondary battery in response to the second state of health being greater than a second threshold.

Abstract: A method and a device for multi-object tracking, and an electronic device are provided. The method includes: determining a hybrid-time position map of a current point cloud fragment; converting a tracked position map of a previous point cloud fragment into a temporary tracked position map of the current point cloud fragment; and averaging the hybrid-time position map and the temporary tracked position map of the current point cloud fragment, to generate a tracked position map of the current point cloud fragment. With the method and the device for multi-object tracking, and the electronic device, the hybrid-time position map and temporary tracked position map of the current point cloud fragment are averaged, so that not only the tracked position map of the current point cloud fragment is accurately generated, but also an object ID is inherited.

Abstract: A method and a device for self-supervised learning, a storage medium, and an electronic device are provided. The method includes: organizing real points in one column along a vertical direction into a pillar, where the pillar is provided with a pillar motion parameter, and each of the real points in the pillar has a motion parameter that is the same as the pillar motion parameter; determining a predicted point in a next frame; determining a first loss term based on a minimum distance among distances between predicted points in the next frame and real points in the next frame, and generating a loss function including the first loss term; and performing self-supervised learning processing based on the loss function. With the method and device for self-supervised learning, the storage medium, and the electronic device, a pillar motion parameter representing motion of a real point is determined with the pillar as a unit, so as to enhance correlation between point clouds.

Abstract: Systems and methods for predicting a location of a target vehicle are disclosed. A processor receives trajectory information for the target vehicle and determines a trajectory feature based on the trajectory information. The processor further determines a lane feature of a lane within a threshold vicinity of the target vehicle, and determines a probability associated with the lane based on the trajectory feature and the lane feature. The lane that the vehicle may enter is identified based on the probability. The processor may also generate interactive features of interactions between the target vehicle and other vehicles. The processor predicts a location of the target vehicle in an upcoming time period based on the lane feature and the trajectory feature. In some embodiments, the interactive features are also used for the prediction.

Abstract: Described herein are systems and methods for jointly learning geometry and motion with three-dimensional holistic understanding. In embodiments, such approaches enforce the inherent geometrical consistency during the learning process, yielding improved results for both tasks. In embodiments, three parallel networks are adopted to predict the camera motion (e.g., MotionNet), dense depth map (e.g., DepthNet), and per-pixel optical flow between consecutive frames (e.g., FlowNet), respectively. The information of 2D flow, camera pose, and depth maps, are fed into a holistic 3D motion parser (HMP) to disentangle and recover per-pixel 3D motion of both rigid background and moving objects. Various loss terms are formulated to jointly supervise the three networks. Embodiments of an efficient iterative training strategy are disclosed for better performance and more efficient convergence.

Abstract: Described herein are systems and methods for jointly learning geometry and motion with three-dimensional holistic understanding. In embodiments, such approaches enforce the inherent geometrical consistency during the learning process, yielding improved results for both tasks. In embodiments, three parallel networks are adopted to predict the camera motion (e.g., MotionNet), dense depth map (e.g., DepthNet), and per-pixel optical flow between consecutive frames (e.g., FlowNet), respectively. The information of 2D flow, camera pose, and depth maps, are fed into a holistic 3D motion parser (HMP) to disentangle and recover per-pixel 3D motion of both rigid background and moving objects. Various loss terms are formulated to jointly supervise the three networks. Embodiments of an efficient iterative training strategy are disclosed for better performance and more efficient convergence.