Physical Sensors & Devices

Sensing, actuation and imaging applications based on ultrasounds could expand to many applications by means of miniaturization and low power consumption via MEMS fabrication technologies. Piezoelectric micromachined ultrasound transducers (PMUTs) with thin film designs have emerged as key commercial products but current state-of-art PMUTs are limited by the acoustic power/pressure for applications within a limited range by using AlN as the piezoelectrical material due to its process compatibility with microelectronics. One BSAC industrial member has developed a process to make PMUT devices...

Space exploration often costs multiple millions of dollars for each exploratory mission to get a single piece of equipment into orbit. These missions usually return information in the form of scans or images or samples in the form of extracted material. This work proposes the manufacture and deployment of thousands of imaging capable solar sails systems with 10 gram payloads. Power generation is enabled through solar panels and batteries. Navigation is enabled through one square meter solar sails maneuvered by inchworm motors. Communications are enabled by laser transmitters and SPAD...

We present a surface-tension driven electrochemical liquid metal (LM) actuator without the gas-producing sidereaction and capable of fabrication/operation in ambient air for practical applications. A hybrid supercapacitor is introduced to inhibit the common counter electrode side reactions, and the use of quasi-solid-state ionic hydrogel instead of liquid electrolyte further enables non-volatile operations. A 2×4 LM droplet array is demonstrated to actuate by a low driving voltage of 3.5 V for a maximum force of ~8.5 mN and a displacement of 0.56 mm in only 1.75 s. With the favorable...

3D perception is an essential task for autonomous driving, and thus building the most accurate, computationally efficient, fast, and label efficient models is of great interest. In particular, label-efficient 3D detection is attractive as manual labeling of 3D LiDAR point clouds is both costly and time-consuming. Autolabeling is a machine learning paradigm in which a model is trained on a (small) set of labeled data before being used to generate predictions, known as pseudo-labels, on a large set of unlabeled data which can then be used to train an accurate downstream model with only a...