3D automatic annotation has received increased attention since manually annotating 3D point clouds is laborious. However, existing methods are usually complicated, e.g., pipelined training for 3D foreground/background segmentation, cylindrical object proposals, and point completion. Furthermore, they often overlook the inter-object feature correlation that is particularly informative to hard samples for 3D annotation. To this end, we propose a simple yet effective end-to-end Context-Aware Transformer (CAT) as an automated 3D-box labeler to generate precise 3D box annotations from 2D boxes, trained with a small number of human annotations. We adopt the general encoder-decoder architecture, where the CAT encoder consists of an intra-object encoder (local) and an inter-object encoder (global), performing self-attention along the sequence and batch dimensions, respectively. The former models intra-object interactions among points and the latter extracts feature relations among different objects, thus boosting scene-level understanding. Via local and global encoders, CAT can generate high-quality 3D box annotations with a streamlined workflow, allowing it to outperform existing state-of-the-arts by up to 1.79% 3D AP on the hard task of the KITTI test set.

Manually annotating 3D point clouds is laborious and costly, limiting the training data preparation for deep learning in real-world object detection. While a few previous studies tried to automatically generate 3D bounding boxes from weak labels such as 2D boxes, the quality is sub-optimal compared to human annotators. This work proposes a novel autolabeler, called multimodal attention point generator (MAP-Gen), that generates high-quality 3D labels from weak 2D boxes. It leverages dense image information to tackle the sparsity issue of 3D point clouds, thus improving label quality. For each 2D pixel, MAP-Gen predicts its corresponding 3D coordinates by referencing context points based on their 2D semantic or geometric relationships. The generated 3D points densify the original sparse point clouds, followed by an encoder to regress 3D bounding boxes. Using MAP-Gen, object detection networks that are weakly supervised by 2D boxes can achieve 94 ~ 99% performance of those fully supervised by 3D annotations. It is hopeful this newly proposed MAP-Gen autolabeling flow can shed new light on utilizing multimodal information for enriching sparse point clouds.

There exist many types and configurations of switched-capacitor converters. However, they are mainly designed for specific applications such as DC-DC or DC-AC (multilevel converters) power conversions. Consequently, they are inapplicable in situations that require multi-applications. A better alternative would be to develop SC converters that are of multi-functions. Different types of power conversions can be achieved by a generalized SC converter, and is possible to fabricate the proposed converter into an IC package. The manufacturing cost can be further reduced through a mass production process. Nevertheless, fundamental studies and clarifications on the feasibility of the idea are still required as relevant literature are not reported. In this work, a preliminary study on a universal SC converter is presented. By programming the control sequence of power switches, the proposed converter is capable of achieving DC-DC, AC-DC, and DC-AC power conversions, while having the properties of bi-directional power flow and variable conversion ratios. The detailed explanations of its circuit operations, control methodology, and basic design procedures, are discussed. Experimental results are included to verify the feasibility of the proposed converter for low voltage applications.

In this paper, a two-layer adaptive differential evolution (ADE) algorithm is adopted to monitor the parameters of the receiving resonators and the mutual inductances of series-series (SS)-compensated wireless power transfer (WPT) systems. By only measuring the primary coils' voltages and currents, the proposed monitoring method can be applied for multiple-coil SS-compensated WPT systems without any feedback signals from the receivers. Compared to the conventional monitoring method based on the genetic algorithm (GA), which may find local optimal solutions by the manually tuned parameters of the mutation rate, the crossover rate, and the generations, the proposed method based on the two-layer ADE can always find global optimal solutions by the automatically tuned parameters of the differential weight, the crossover rate, and the generations. Experimental results validate that the ADE and the proposed two-layer ADE can monitor the parameters of both two- and three-coil SS-compensated WPT systems more steadily and accurately than the conventional GA. Additionally, the proposed two-layer ADE is verified to monitor the parameters of three-coil SS-compensated WPT systems with three different arrangements more accurately than the ADE.

There is a demand for low-power high-voltage-gain transformerless DC-DC converters. It is difficult to achieve this kind of conversion with good efficiency through a buck converter. The two-stage converter is proven to be an effective solution for such applications. In this paper, a series of exponential step-down switched-capacitor (ESC) converters that achieve the benefits of high-voltage-gain conversion, but comprise fewer switches than other switched-capacitor converter topologies, is proposed. The ESC converters are applicable as first-stage converters in the two-stage power conversion solution. Experimental results validating the idea are provided.

This paper presents an analysis on the effect of having different number of capacitors n in the first-stage switched-capacitor circuit of an improved hybrid switched-capacitor buck converter for high-voltage-gain conversion. Various aspects of the topology, operation, and efficiency are investigated. It is shown that a higher n in the step-down capacitor stage does not necessarily lead to an overall improved power efficiency.

The energy-efficiency issue of the switched-capacitor converters is still a highly controversial topic that requires a more in-depth exploration. This paper will address the issue by dissecting the analysis of the entire efficiency problem into two parts. In the first part, the efficiency analysis of charging the capacitor of an RC circuit under different aspects (partial charging, full charging, at zero capacitor voltage, at non-zero capacitor voltage, etc.) will be conducted. In the second part, the efficiency analysis of discharging the capacitor of an RC circuit with resistive and capacitive loads will be covered. A complete evaluation of the overall efficiency is then performed in terms of both the charging and discharging efficiencies of the capacitor. Additionally, it is shown in this paper that the claim that quasi-switched-capacitor converters are more lossy than switched-capacitor converters is a common misconception.

The switched-capacitor converters are ideal switching-mode power supplies for portable electronic consumers due to their light weight, small size, and high power density. However, they suffer from a discontinuous input current waveform with large di/dt, what leads to significant electromagnetic interference (EMI) emission. This paper proposes a configuration of switched-capacitor converters connected in parallel with their inputs and outputs interleaved. The interleaving times are calculated by taking into account the fast capacitor-charging characteristic, and the need to have the nominal operating point on its linear part for increasing the regulation range at changes in the input voltage and load. To improve control performance and avoid the use of small-signal linearization in the control design, a large-signal approach is adopted for modeling the converter, allowing for a design of a sliding mode control.

This paper discusses a family of fixed-frequency pulse-width-modulation based sliding mode voltage controllers for basic DC-DC converters operating in continuous conduction mode. An experimental comparison is offered between the proposed controller and the PWM peak current mode controller for the case of the boost converter. The performances and properties of the proposed controller, the conventional PWM voltage mode controller, and the PWM current mode controllers are also compared.