Search

A reliable, efficient, and simple control is presented and validated for a quarter-car active suspension system equipped with an electro-hydraulic actuator. Unlike the existing techniques, this control does not use any function approximation, e.g., neural networks (NNs) or fuzzy-logic systems (FLSs), while the unmolded dynamics, including the hydraulic actuator behavior, can be accommodated effectively. Hence, the heavy computational costs and tedious parameter tuning phase can be remedied. Moreover, both the transient and steady-state suspension performance can be retained by incorporating prescribed performance functions (PPFs) into the control implementation. This guaranteed performance is particularly useful for guaranteeing the safe operation of suspension systems. Apart from theoretical studies, some practical considerations of control implementation and several parameter tuning guidelines are suggested. Experimental results based on a practical quarter-car active suspension test-rig demonstrate that this control can obtain a superior performance and has better computational efficiency over several other control methods. [ABSTRACT FROM AUTHOR]

Feature matching is a crucial technique to estimate robot pose in the light of natural landmarks for a vision navigation system. It is difficult for the present feature matching approaches to balance the accuracy, robustness and efficiency for a vision-based robot in complex workspaces. A multi-stage KNN-TF-BF-EN-RANSAC (KTBER) refinement matching technique is proposed to combine with the adaptive ORB feature extraction method, in order to develop a novel KTBER_AORB model that can outperform the state-of-the-art feature extraction and matching techniques. Firstly, the adaptive ORB (AORB) features are extracted in the Laplacian Transformation of Gaussian (LTOG) pyramid by using a set of variable-radius templates. The template radius is adjusted in terms of different brightness and contrast conditions. Then, the Eigenvector Norm (EN) and Bidirectional Filtering (BF) techniques are devised to establish a novel KTBER model, which can match AORB feature pairs not only according to the local appearance-based similarity of feature descriptors in the Hamming space, but also by means of the similarity of local image structure and global grayscale information for feature pairs in the Euclidean space. Finally, a large number of feature matching experiments are conducted on five public datasets and our mobile robot dataset, in order to test the algorithm performances for different challenging scenarios, including illumination change, motion blur, image rotation, viewpoint change, and low texture. On the one hand, an intensive experimental analysis of our model with different configurations is made to check the specific effects of different algorithm elements. On the other hand, the KTBER_AORB model is further compared with four state-of-the-art techniques, e.g., SIFT, ORB-SLAM, MP-ORB, and LAAT. The experimental results demonstrate that the KTBER_AORB model outperforms other state-of-the-art feature matching techniques on accuracy and robustness, still preserving a higher efficiency than most counterparts. [ABSTRACT FROM AUTHOR]

With the rapid development of Internet-of-Things (IoT), self-powered visual perception systems attract lots of attention. This brief presents a near-sensor motion detection architecture for the energy harvesting image sensor targeting always-on monitoring applications. Reconfigurable pixel structures are designed to harvest solar energy from the focal plane and conduct motion detection simultaneously. The near-sensor motion detection architecture combining analog-domain edge detection and digital-domain frame differencing technique is proposed to reduce the power consumption on data conversion significantly. A prototype chip of 32 × 32 array size is fabricated with the standard 0.18 μm CMOS process. The measurement results show that the chip prototype achieves motion detection at a frame rate of 1894 fps consuming 16.43 μW power. Furthermore, the chip harvests 1.263 μW power under 100kLux illumination, and the analytical computation result shows that self-powered motion detection can be achieved at a frame rate of 58 fps. [ABSTRACT FROM AUTHOR]

Intelligent systems have brought convenience to contemporary society. However, latency and poor-efficiency have been urgent problems for intelligence systems. Here, an infrared (IR) intelligent system fusing a non-contact IR thermopile sensor array fabricated by microelectromechanical system technology and an artificial neural network (ANN) algorithm is proposed, with the characteristics of high-efficiency and low-latency. The system is based on a designed near-sensor computing architecture, which can realize computing directly on edge devices without sending data to the cloud, resulting in reduced redundant data and decreased latency. The responsivity of the sensors affects the weight accuracy and computing speed of the ANN algorithm and further influences the accuracy and efficiency of the system. Transfer of graphene oxide (GO) material to the IR thermopile sensors with a proposed location transfer method is suggested to enhance the responsivity. The responsivity of the IR thermopiles with GO is up to 705.1 V/W, which is enhanced by 85.9% compared to that without GO. The system is applied to gesture recognition to study practicality. The recognition accuracy of the system with GO is 100%. This work provides an effective idea for studying a high-accuracy IR intelligent sensing system. [ABSTRACT FROM AUTHOR]