UB0100 CubeSat Board for AI and Computer Vision Acceleration

The Power of Myriad 2 brought to Cubesats

The Ubotica UB0100 CubeSat Board brings the power of Myriad 2’s CV and AI compute acceleration to a PC104 form-factor for smallsat and cubesat missions. It is built around the Intel® Movidius™ Myriad™ 2 Computer Vision (CV) and Artificial Intelligence (AI) COTS VPU whose 12 vector cores provide high-performance compute within a low power envelope (nominally < 1.5W).

Myriad 2 onboard PhiSat-1

The Myriad 2 is currently undergoing complete SEE radiation characterisation in conjunction with ESA, and in parallel will fly onboard the PhiSat-1 cubesat that is due for launch in June 2020, providing early device flight data and space heritage.

The UB0100 is the next generation of the Myriad board that is flying on PhiSat-1, and will be available from Q3 2020.

Flexible & Easy to Integrate

UB0100_functional_block_diagram

The UB0100 combines the power efficient compute of the Myriad 2 VPU with a wide range of interfaces and peripherals, providing flexibility for integration into satellite platforms.

Either Gigabit Ethernet or USB2.0/3.0 can be used as the primary control and data interface to the board, enabling data rates sufficient to handle many CV and AI applications at near-streaming throughput.

Power distribution and latch-up protection is managed on-board, requiring only a single 5V external source.

Widely Adaptable for AI and Vision Deployments

Flexible Operational Modes

In Direct Mode, a master host communicates with the UB0100 to load CV and NN models to the Myriad, and to thereafter submit image frames to the board and receive processing results. The host can dynamically update the Myriad application over the primary interface, enabling flexible runtime solutions.

In Custom Mode, the board can be controlled over CAN or SPI, in addition to Ethernet and USB. The capture and processing of image frames from a directly connected image sensor over a standard 16-lane parallel interface is facilitated in this mode.

Neural Network Inference

Common NN frameworks (e.g., TensorFlow, PyTorch, Caffe) can be used for NN model development and training, with the model subsequently imported into Intel®’s OpenVINO™ toolkit for targeting to the Myriad device. The UB0100 leverages the broad range of pre-qualified models and layers available within OpenVINO. Pre-trained models available in the OpenVINO application zoo can be used with transfer learning, or can be deployed directly on the UB0100 board, via the CVAI Toolkit software, without requiring any modifications.

CV and ISP Pipelines

Custom Computer Vision pipelines can easily be deployed and executed on the UB0100 using the Ubotica CVAI Toolkit™ software*. The software supports the graphical design of application-specific CV and ISP pipelines that utilise a combination of the power-efficient streaming hardware blocks within the Myriad 2 and software filters implemented on the vector engines. Deployment to the UB0100 involves the transfer of only a single configuration file. Runtime updates enable the switching and tuning of pipelines without requiring application re-compile or system reboot.

Compact & Self Contained

Boot can be either over USB or SPI from an external device, or from an on-board 1GB NOR flash. Radiation and inertial sensing components, as well as non-volatile data storage, are also provided on the board.

The compact 94mm x 91mm PC104 form factor is suitable for integration into 1U cubesats.

Features

The UB0100 board has the following features**:

  • MA2450 Myriad 2 CV and AI engine for smallsat and cubesat applications

  • PC104 form factor, meeting ESA’s board specification for cubesat demonstrators

  • On-board integrated latch-up protection

  • Operates from single 5V supply

  • Direct operation mode → CV and AI engine as a slave over USB 2.0/3.0 or Ethernet
  • Custom operation mode → CV and AI for directly interfaced sensors, control over SPI, CAN, USB or Ethernet (UART and I2C also available)

  • COTS components

  • Serial NOR flash for boot

  • 3.3V I/O voltage

  • Parallel and MIPI camera interfaces for direct sensor connection

  • Compatible with Ubotica CVAI Toolkit Software, for complete control over CV and NN processing

  • microSD card for non-volatile data storage

  • Additional integrated satellite-oriented peripherals

    • Floating Gate Dosimeter for radiation sensing
    • IMU integrating 3D accelerometer and gyroscope

  • Targets sub 5W power envelope (application dependent)

  • Assembly-time reconfigurability via 0-ohm resistors

  • Stackable 2x dual row board-edge headers for power delivery and platform/payload communications

  • High-speed parallel camera interface for direct sensor interfacing

  • Full product sheet and board availability Q3 2020

* In Direct Mode

** This board is being developed under an ongoing ESA-supported project, and the information provided here is subject to modification.