Guided Depth Completion with Instance Segmentation Fusion in Autonomous Driving Applications.

Pixel-level depth information is crucial to many applications, such as autonomous driving, robotics navigation, 3D scene reconstruction, and augmented reality. However, depth information, which is usually acquired by sensors such as LiDAR, is sparse. Depth completion is a process that predicts missing pixels' depth information from a set of sparse depth measurements. Most of the ongoing research applies deep neural networks on the entire sparse depth map and camera scene without utilizing any information about the available objects, which results in more complex and resource-demanding networks. In this work, we propose to use image instance segmentation to detect objects of interest with pixel-level locations, along with sparse depth data, to support depth completion. The framework utilizes a two-branch encoder-decoder deep neural network. It fuses information about scene available objects, such as objects' type and pixel-level location, LiDAR, and RGB camera, to predict dense accurate depth maps. Experimental results on the KITTI dataset showed faster training and improved prediction accuracy. The proposed method reaches a convergence state faster and surpasses the baseline model in all evaluation metrics.

Highly Sensitive Porous PDMS-Based Capacitive Pressure Sensors Fabricated on Fabric Platform for Wearable Applications.

A novel porous polydimethylsiloxane (PDMS)-based capacitive pressure sensor was fabricated by optimizing the dielectric layer porosity for wide-range pressure sensing applications in the sports field. The pressure sensor consists of a porous PDMS dielectric layer and two fabric-based conductive electrodes. The porous PDMS dielectric layer was fabricated by introducing nitric acid (HNO3) into a mixture of PDMS and sodium hydrogen bicarbonate (NaHCO3) to facilitate the liberation of carbon dioxide (CO2) gas, which induces the creation of porous microstructures within the PDMS dielectric layer. Nine different pressure sensors (PS1, PS2,..., PS9) were fabricated in which the porosity (pore size, thickness) and dielectric constant of the PDMS dielectric layers were varied by changing the curing temperature, the mixing proportions of the HNO3/PDMS concentration, and the PDMS mixing ratio. The response of the fabricated pressure sensors was investigated for the applied pressures ranging from 0 to 1000 kPa. A relative capacitance change of ∼100, ∼323, and ∼485% was obtained by increasing the curing temperature from 110 to 140 to 170 °C, respectively. Similarly, a relative capacitance change of ∼170, ∼282, and ∼323% was obtained by increasing the HNO3/PDMS concentration from 10 to 15 to 20%, respectively. In addition, a relative capacitance change of ∼94, ∼323, and ∼460% was obtained by increasing the PDMS elastomer base/curing agent ratio from 5:1 to 10:1 to 15:1, respectively. PS9 exhibited the highest sensitivity over a wide pressure sensing range (low-pressure ranges (<50 Pa), 0.3 kPa-1; high-pressure ranges (0.2-1 MPa), 3.2 MPa-1). From the results, it was observed that the pressure sensors with dielectric layers prepared at relatively higher curing temperatures, higher HNO3 concentrations, and higher PDMS ratios resulted in increased porosity and provided the highest sensitivity. As an application demonstrator, a wearable fit cap was developed using an array of 16 pressure sensors for measuring and mapping the applied pressures on a player's head while wearing a helmet. The pressure mapping aids in observing and understanding the proper fit of the helmet in sports applications.

A highly sensitive printed humidity sensor based on a functionalized MWCNT/HEC composite for flexible electronics application.

A novel functionalized multi-walled carbon nanotube (FMWCNT)/hydroxyethyl cellulose (HEC) composite-based humidity sensor was successfully developed for humidity monitoring applications. FMWCNTs were synthesized by covalently functionalizing multi-walled carbon nanotubes (MWCNTs) in a mixture of sulfuric and nitric acid to enhance their hydrophilicity. The FMWCNTs were characterized using transmission electron microscopy, Raman spectroscopy, Fourier transform infrared spectroscopy and dispersion analysis to verify the presence of functional hydroxyl and carboxyl groups. A FMWCNT/HEC (1 : 6 w/w) composite ink was formulated using the solution blending technique with 2.5 wt% FMWCNTs. A multi-layered humidity sensor was fabricated using additive print manufacturing processes on a flexible polyethylene terephthalate (PET) substrate. Screen printing and gravure printing processes were used to deposit the bottom silver (Ag) electrode and FWMCNT/HEC sensing layers, respectively. The capability of the fabricated humidity sensor was investigated by measuring its resistive response towards relative humidity (RH) varying from 20% RH to 80% RH. As the RH was increased from 20% RH to 80% RH in steps of 10% RH at 25 °C, it was observed that the resistance of the printed sensor increased linearly. The printed sensor demonstrated resistance changes as high as ≈290% at 80% RH, when compared to its base resistance at 20% RH. A sensitivity and a response time of 0.048/%RH and ≈20 s were obtained for the printed sensor, respectively. The results thus demonstrated the feasibility of employing additive print manufacturing processes to develop a highly sensitive sensor for humidity monitoring applications.