Embedded Archives - Page 50 of 92 - SmartCitiesElectronics.com

Sensor software library supports multiple dynamic objects

Fri, Mar 11, 2022

Sensor fusion specialist, Baselabs, has released the latest version of its modular, ISO 26262-certified software library for the development of data fusion systems for automated driving functions. Baselabs Create Embedded 8 supports distinct models for different types of dynamic objects in its software development kit (SDK).

Developers of advanced driver assistance systems (ADAS) control units can obtain, for example, a constant velocity model for pedestrians and a constant curvature model for vehicles. If an object is not yet assigned to a class, for example during initialisation, multiple hypotheses with different models can be created. Multiple-model support increases the reliability of predicting movements of different object classes, claimed Baselabs. As a result, the measurement association and the overall performance of sensor fusion are significantly improved.

The Baselabs Create Embedded data fusion library provides sensor fusion algorithms for combining data from radar, camera, and lidar sensors. Developers can configure the data fusion of sensors from a range of manufacturers with graphical support and generate a consistent object list of the vehicle environment. This list serves as an input for downstream algorithms and ADAS functions.

The library offers self-contained C code for engine control units (ECUs) such as Infineon Aurix or Arm Cortex, for example, and is safety-certified to ASIL B.

The height models can be pre-set and there is also the option to use categorical values to better separate objects that are close to each other.

The software reduces maintenance and allows teams to focus on other design areas, said Robin Schubert, CEO of Baselabs. “By licensing the library, development teams can focus on designing differentiating features and reduce time-to-market for their applications. Maintenance efforts for in-house developed software are eliminated.”

Baselabs provides software products to make the development of sensor fusion efficient and scalable. It is partner for automotive manufacturers and suppliers worldwide. The company was founded in 2012 and is in shared ownership by its four founders and Vector Informatik and is thus strategically independent of any OEM, Tier 1, or sensor supplier in the market

https://www.baselabs.de/object-fusion/baselabs-create-embedded

> Read More

GaN FETs perform for e-bikes, drones and robotics, says EPC

Thu, Feb 24, 2022

Introducing the EPC9167 GaN-based inverter reference design, Efficient Power Conversion (EPC) said that the FET enhances motor system performance, range, precision and torque, while lowering overall system cost.

The three-phase brushless DC (BLDC) motor drive inverter uses the EPC2065 eGaN FET. The EPC9167 operates from an input supply voltage between 14 and 60V (nominal 48V) and is available in two configurations – a standard unit and a high current version.

The EPC9167 standard reference design board is a three-phase BLDC motor drive inverter board featuring the EPC2065 eGaN FET rated at 3.6 mOhm maximum RDS(on), 80V maximum device voltage. This standard configuration uses single FETs for each switch position and can deliver up to 20A RMS maximum output current.

For the high current configuration, the EPC9167HC, reference design uses two paralleled FETs per switch position with the ability to deliver up to 42Apk (30A RMS) maximum output current.

Both versions contain all the necessary critical function circuits to support a complete motor drive inverter including gate drivers, regulated auxiliary power rails for housekeeping supplies, voltage, and temperature sense, accurate current sense, and protection functions. The boards feature the ST Microelectronics’ STDRIVEG600, smart motor drive GaN half-bridge driver.

The EPC9167 boards measure just 130 x 100mm (including the connector). The boards can also be configured for multi-phase DC/DC conversion and support both phase and leg shunt current sensing. The boards demonstrate benefits of a GaN-based motor drive, for example lower distortion for lower acoustic noise, lower current ripple for reduced magnetic loss, lower torque ripple for improved precision, and lower filtering for lower cost.

The EPC9167 boards’ low weight and small size enable them to be integrated into the drive into the motor housing and support low inductance, higher power density motors.

EPC provides full demonstration kits, which include interface boards that connect the inverter board to the controller board development tool for fast prototyping.

The default setting for the GaN-based motor drive kit is 100kHz switching frequency and 14 ns deadtime. The kit is designed to be programmed for different frequencies and deadtimes, although operation at around 100kHz eliminates the need for electrolytic capacitors and allows the use of lower capacitance. It also reduces motor losses, said EPC. The company also advised operating the boards with very small low deadtime of around 14 ns to allow higher torque per Ampere. Combined these features improve inverter and motor system efficiency by more than seven per cent compared with a silicon MOSFET, which typically operates at 20kHz and 500 ns deadtime, said EPC.

The EPC9167 and the EPC9167HC reference design are available for immediate delivery from Digi-Key.

http://www.epc-co.com

> Read More

CompuRAM technology enables AI at the edge capacity

Wed, Feb 09, 2022

Embedded memory specialist, sureCore has developed CompuRAM in-memory computing technology for power efficient IoT computing. Sensor data often has to be sent from an IoT device to a server for processing, which creates a connectivity requirement and an unavoidable latency, explained sureCore. This is not acceptable for time critical applications and any extra AI-related computation must be done in a power-efficient way, said CompuRAM.

sureCore’s existing low-power memories provide a way to add the significant extra memory needed by AI applications without dramatically increasing power requirements, said the company. In-memory computing provides further power savings, it continued, by reducing the need to move large amounts of data around within a chip, as the initial processing of data is carried out very close to the memory array itself.

Chief technology officer, Tony Stansfield, explained: “sureCore’s in-memory compute technology achieves this integration by embedding arithmetic capability deep within the memory array in a way that is compatible with its existing silicon-proven, low-power memory design”

sureCore is a low power, embedded memory specialist, which empowers the IC design community to meet aggressive power budgets through a portfolio of low power memory design services and standard IP products. sureCore says its low-power engineering methodologies and design flows meet the most exacting memory requirements with a comprehensive product and design services that create clear market differentiation for customers. The company’s low-power product line encompasses a range of close to near-threshold, silicon proven, process-independent SRAM IP.

http://www.sure-core.com

> Read More

Axiomtek’s RSC100 edge AI computer is available from Impulse Embedded

Mon, Feb 07, 2022

Axiomtek’s RSC100, ARM-based Edge embedded PC featuring the Hailo-8 AI accelerator is now available from industrial computing systems specialist, Impulse Embedded. The streamlined processing unit for AI applications can offer up to 26 T operations per second and is housed in a cost-effective, power-efficient unit, said the company.

The edge system combined with Hailo’s Hailo-8 processor, can be used in a range of applications, including public safety, smart factory, agriculture and intelligent transportation.

The system chassis is made up of aluminium and heavy duty steel with an IP40 rating and supports a wide operating temperature of -20 to +70 degrees C, making it suitable for use in harsh, industrial environments where the temperatures can vary throughout operation. The RSC100 has dual Gigabit Ethernet ports, an M.2 3052 B-key slot which can be used to install a 5G module and two full-size miniPCIe slots. The RSC100 also has seven SMA antenna breakout holes for wireless comms requirements of a variety of applications.

The eight-core ARM processor is backed up with 4Gbyte of LPDDR4 system memory and has 16Gbyte of eMMC storage onboard as standard as well as an M.2 2280 M-key SSD slot with PCIe x4 NVMe support and a MicroSD card slot for additional storage.

Further I/O includes a built-in HDMI 2.0 port with 4K resolution support, two serial ports supporting RS-232/422/485, two CANbus ports, two USB2.0 ports and eight channels of digital I/O, (four, direct input, four direct output). For embedded operating system support, Yocto Linux 3.0 is supplied.

The I/O, software and connectivity, combined with an AI-focused microprocessor that uses just 2.5W whilst achieving 26TOPs of INT-8 performance, means the RSC100 can process complex deep learning neural networks at the edge in smart embedded applications, explained Impulse Embedded.

As with all of its embedded computing range, Impulse can configure the RSC100 to customer’s exact specifications in its UK-based engineering facility. It offers a choice of storage, peripheral cards, operating system, and neural network.

The company has embedded systems capabilities and a team of in-house engineers and specialists to create revision-controlled systems which can reduce project costs and development time.

http://www.impulse-embedded.co.uk

> Read More

Sensor software library supports multiple dynamic objects

GaN FETs perform for e-bikes, drones and robotics, says EPC

CompuRAM technology enables AI at the edge capacity

Axiomtek’s RSC100 edge AI computer is available from Impulse Embedded

About Smart Cities