The development of autonomous unmanned aerial vehicles has been the subject of intensive research for past decade. The introduction of Ardupliot, an open source autopilot software suite, encouraged hobbyists as well as developers to start experimenting aerial vehicles for autonomous missions such as land survey, aerial mapping, etc. In this study, we have attempted to interface a companion PC (Raspberry Pi) with an autopilot system (Pixhawk Mini 3DR) using Python programming and Ardupilot software and establishing short range telemetry communication between the two using Bluetooth modules. The use of RF telemetry modules is quite expensive and unnecessary during indoor testing. It also makes the drone cumbersome to build and reduces flight time owing to the larger current draw of the telemetry modules. Hence in this study, we have used HC-05 modules to successfully transmit and receive data and were also able to use mobile application of Q Ground Control to perform remote operations such as arming, changing of modes, etc. The setup is highly recommended for testing image processing algorithms using the Raspberry Pi (RPi). The study also aims at giving a detailed account of the connection, setup and programming of a 250mm mini drone using RPi. In this project, we have developed an image processing algorithm, successfully interfaced RPi with Pixhawk Mini autopilot using Bluetooth telemetry and were able to conduct a test flight that demonstrates autonomous navigation and landing of the prototype under study.
The advancement in biomedical engineering leads to the development of many technologies in the field of neurology. Researches are going on the study of analysis of autism using electroencephalogram (EEG) signals. EEG is preferred because it is non invasive and it contain information regarding the brain function. This work aims to investigate the statistical difference in EEG signal between autistic and normal. The EEG signals are decomposed into different sub-bands using discrete wavelet transform. The statistical features like mean, standard deviation and power spectral density are used to extract the features from each sub-band. The results show that standard deviation and power spectral density plays an important role in analyzing the EEG signal in sleep, awake and photic conditions.
Reconfigurable antenna operating at 2.45 GHz for Internet of Things (IoT) applications is presented. The antenna is made with concentric circular shaped discs connected with six PIN diodes to reconfigure the operating frequency and the radiation pattern. The proposed antenna is fed with coaxial feed at center position. The ground plane of the antenna is made with circular disc and circular ring. Rogers 3006 flexible material with dielectric constant 6.15 is used as substrate. The reconfigurable antenna simulated analysis is presented with various diode switching conditions with the help of reflection coefficient and the radiation patterns.
This study aims to report the fan-out-four (FO4) delay and energy consumption associated with the various circuit used in the implementation of the neural network signal processing using the 7nm node. This study characterizes the behavior of the 7nm-FinFET to design the various logic circuits and then report the FO4 delay and the energy consumption associated with the basic gates, 16-bit Brent-Kung adder, Flip-flop, 16-bit array pipeline multiplier (clocked at 4GHz). The successful design of the 6T Static Random-Access memory (SRAM) and the clocked regenerative latch-based comparator (clocked at 4GHz) shows that the high-speed memory and mixed-signal circuit design is possible with 7nm FinFET. For the simulation, 7nm FinFET predictive Technology model (PTM) for bulk has been used with the BSIMCMG model. Device characterization shows that the drain current is proportional to the number of fins (NFIN). During the simulation, it is observed that the threshold voltage (0.3V) and the drain current of the N/P FinFET are almost equal. The measured leakage current is 3.5pA. The comparison of the simulation results with 45nm shows that the 7nm FinFET based logic design is highly efficient in terms of FO4 delay, power, and energy. Despite the reduced supply voltage in the 7nm node, the noise margin of the SRAM is still comparable to 45nm.
Matching is an essential operation towards automated registration of digital images. Registration of images is basically done either by studying the characteristics of the region surrounding the object of interest or by considering image morphological features. Many automated matching algorithms have been proposed and evaluated in the literature. However, previous studies on automated matching have either used a small number of benchmarks images in their evaluation, been directed at microscopy and not digital cameras, or have been based on ad hoc evaluation criteria. In this paper, we are presenting an image independent automated algorithm for pixel-to-pixel matching between two digital images obtained from two different microscope imaging devices. Here, SURF feature points and Euclidean distance measures are used for identifying key matching member index (KM2I). We have extensively tested our algorithm on a macro set of 1,69,000 digital images obtained from 250 pap smear samples. Experimental results prove that the proposed algorithm performs better for digital photography.
