Abstract:
A conventional flow tube for a metabolic cart is usually a straight length of pipe whose inner diameter is fixed by the respiratory burden imposed by the flow tube on the user, with a smaller diameter imposing a higher respiratory burden. The ratio of the straight flow tube's length to diameter is fixed by fluid dynamics, so increasing the flow tube's diameter causes the flow tube's length to increase. As the flow tube gets longer, it exerts more torque on the user's neck and jaw, creating discomfort. Reducing the flow tube's length causes an undesired increase in the respiratory burden but increasing the flow tube's diameter to reduce the respiratory burden makes the flow tube less comfortable, making the flow tube unconformable, hard to breathe through, or both. Bending the flow tube, e.g., in an L shape, makes it possible to increase the flow tube's propagation length without increasing the flow tube's lever arm length.
Abstract:
The Zeeman shift of electronic spins in nitrogen-vacancy (NV) centers in diamond has been exploited in lab-scale instruments for ultra-high-resolution, vector-based magnetic sensing. A quantum magnetometer in CMOS utilizing a diamond-nanocrystal layer with NVs or NV-doped bulk diamond on a chip-integrated system provides vector-based magnetic sensing in a compact package. The system performs two functions for the quantum magnetometry: (1) strong generation and efficient delivery of microwave for quantum-state control and (2) optical filtering/detection of spin-dependent fluorescence for quantum-state readout. The microwave delivery can be accomplished with a loop inductor or array of wires integrated into the chip below the nanodiamond layer or diamond. And the wire array can also suppress excitation light using a combination of plasm onic and (optionally) Talbot effects.
Abstract:
A thermoelectric device includes a thermoelectrode characterized by a band gap less than k B T , where k B is the Boltzmann constant and T is a temperature of the thermoelectrode. The device also includes a magnetic field source, operably coupled to the thermoelectrode, to apply a magnetic field B on the thermoelectrode along a first direction. The device also includes a voltage source, operably coupled to the thermoelectrode, to apply an electric field E on the thermoelectrode along a second direction substantially perpendicular to the first direction so as to generate a heat flow along the second direction.
Abstract:
A thermoelectric device includes a thermoelectrode characterized by a band gap less than k B T , where k B is the Boltzmann constant and T is a temperature of the thermoelectrode. The device also includes a magnetic field source, operably coupled to the thermoelectrode, to apply a magnetic field B on the thermoelectrode along a first direction. The device also includes a voltage source, operably coupled to the thermoelectrode, to apply an electric field E on the thermoelectrode along a second direction substantially perpendicular to the first direction so as to generate a heat flow along the second direction.
Abstract:
A LiDAR device that transmits a single or multiple continuous or intermittent laser beams to the environment and detects the reflected light on one or more detectors. The LiDAR device may include a scanning mirrors array composed of a single or multiple moving mirrors capable of changing the direction of the transmitted light. The scanning mirrors array may also include sensors and actuators which can be used to precisely control or measure the position of the mirrors. The LiDAR device may also include a lens that focuses the light captured by the mirror(s) onto a single or a multitude of detectors. The device may include laser sources and detectors operating in various wavelengths. The LiDAR device may also include laser power modulation mechanisms at a single or multitude of frequencies to improve signal detection performance and remove any ambiguity in range calculation.
Abstract:
Lifelong Deep Neural Network (L-DNN) technology revolutionizes Deep Learning by enabling fast, post-deployment learning without extensive training, heavy computing resources, or massive data storage. It uses a representation-rich, DNN-based subsystem (Module A) with a fast-learning subsystem (Module B) to learn new features quickly without forgetting previously learned features. Compared to a conventional DNN, L-DNN uses much less data to build robust networks, dramatically shorter training time, and learning on-device instead of on servers. It can add new knowledge without re-training or storing data. As a result, an edge device with L-DNN can learn continuously after deployment, eliminating massive costs in data collection and annotation, memory and data storage, and compute power. This fast, local, on-device learning can be used for security, supply chain monitoring, disaster and emergency response, and drone-based inspection of infrastructure and properties, among other applications.
Abstract:
A system and a method for the deterministic generation of magnetic skyrmions includes a magnetic strip configured to store and transport skyrmions. The magnetic strip includes one or more spatial inhomogeneities configured to generate a skyrmion at known locations when excited by a current pulse. A current pulse generator is used to inject current pulses into the magnetic strip via contact pads electrically coupled to both the current pulse generator and the magnetic strip. The system also includes a magnetic field source to apply an out-of-plane magnetic field across the magnetic strip to facilitate generation of skyrmions. Skyrmions can be generated by applying an out-of-plane magnetic field to the magnetic strip and injecting a current pulse with sufficient current density towards the spatial inhomogeneities. Once a skyrmion is generated, another current pulse with sufficient current density can be injected to move the skyrmion.
Abstract:
An imaging system uses a dynamically varying coded mask, such as a spatial light modulator (SLM), to time-encode multiple degrees of freedom of a light field in parallel and a detector and processor to decode the encoded information. The encoded information may be decoded at the pixel level (e.g., with independently modulated counters in each pixel), on a read-out integrated circuit coupled to the detector, or on a circuit external to the detector. For example, the SLM, detector, and processor may create modulation sequences representing a system of linear equations where the variables represent a degree of freedom of the light field that is being sensed. If the number of equations and variables form a fully determined or overdetermined system of linear equations, the system of linear equations' solution can be determined through a matrix inverse. Otherwise, a solution can be determined with compressed sensing reconstruction techniques with the constraint that the signal is sparse in the frequency domain.
Abstract:
An artificial saturable absorber uses additive pulse mode-locking to enable pulse operation of an on-chip laser operation. Four different artificial saturable absorbers are disclosed. The first includes an integrated coupler, two arms each containing some implementation of the end-reflector, and a phase bias element in one arm. The second includes an integrated directional coupler, two integrated waveguide arms, and another integrated coupler as an output. The third includes an integrated birefringent element, integrated birefringent-free waveguide, and integrated polarizer. And the fourth includes a multimode waveguide that allows for different modes to propagate in such a way that the difference in the spatial distribution of intensity causes a nonlinear phase difference between the modes. These are just some examples of an on-chip fully integrated artificial saturable absorber with instantaneous recovery time that allow for generation of sub-femtosecond optical pulses at high repetition rates using passive mode-locking.
Abstract:
An optical neural network is constructed based on photonic integrated circuits to perform neuromorphic computing. In the optical neural network, matrix multiplication is implemented using one or more optical interference units, which can apply an arbitrary weighting matrix multiplication to an array of input optical signals. Nonlinear activation is realized by an optical nonlinearity unit, which can be based on nonlinear optical effects, such as saturable absorption. These calculations are implemented optically, thereby resulting in high calculation speeds and low power consumption in the optical neural network.