CogniSat^TM AI and Computer Vision Payload Processing Platform Specifications

The power of Myriad brought to cubesats

CogniSat^TM AI and Computer Vision Payload Processing Board Specifications

The power of Myriad 2
brought to cubesats

The Ubotica CogniSat^TM 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 parallel and hardware accelerated compute within a low power envelope.

Widely Adaptable for AI and
Computer Vision Deployments

Flexible and Integration Ready

CogniSat^TM combines the power efficient compute of the Myriad 2 VPU with a wide range of interfaces and peripherals, providing broad 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.

Efficient Neural Network Inference

Common NN frameworks (e.g., TensorFlow, PyTorch, Caffe) can be used for NN model development and training, with the model subsequently targeted to CogniSat^TM. CogniSat^TM leverages the broad range of pre-qualified models and layers available within Intel®’s OpenVINO™ toolkit. Pre-trained OpenVINO™ models can be used with transfer learning, or can be deployed directly on CogniSat’s^TM board, via the Ubotica CVAI Toolkit™ software.

Customisable CV and ISP Pipelines

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

Dual Operational Modes

In Direct Mode, a master host communicates with CogniSat^TM 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, frames from a directly connected sensor can be captured, processed and inferenced.

Key Features

CogniSat^TM has the following features**:

PC104 form factor, meeting ESA-TEC’s board specification for cubesat IODs
On-board integrated latch-up protection, with auto-restart and trip signal
Single 5V supply
CAN interface for C&C
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†
All COTS components: Selected based on flight heritage and prior rad test results
Integrated Neural Network Supervisor
Serial NOR boot flash

External power-down control
Parallel and MIPI camera interfaces for direct sensor connection†
Fully compatible with the Ubotica CVAI Toolkit™ Software, for complete control over CV and AI processing
microSD card for non-volatile data storage
Additional on-board peripherals
• Floating Gate Dosimeter for radiation sensing
• IMU (3D accelerometer and gyroscope)
Targets sub 5W power envelope and incorporates power gating
Stackable 2x dual row board-edge headers for power delivery and low-level control

** These features subject to modification
* In Direct Mode
† In Custom Mode

CVAI Toolkit^TM

Ubotica’s CVAI Toolkit^TM is a lightweight library for developers to bring the power of Myriad’s Computer Vision (CV) and Artificial Intelligence (AI) compute acceleration to their in-orbit applications. The full feature set of the Myriad Vision Processing Unit (VPU) is exposed to facilitate the rapid development and deployment of applications to process data on-board satellite in a fast, power efficient manner.

Key Features:

Fully compatible with Myriad Neural Networks developed for Intel’s OpenVINO platform.
Runtime configuration of Myriad’s Hardware and Software CV filters for Image Transformations.
Integrated Neural Network Supervisor
Optimised for environments with a single Myriad VPU
Supports both USB and Ethernet connectivity
APIs for C/C++ and Python programming languages

Interfaces
Ethernet	10/100/1000BaseT
USB	2.0/3.0
CAN	Supporting CSP
SPI	Master/slave, 3.3V logic
I2C	3.3V logic
UART	3.3V logic
GPIO	x5, 3.3V logic
JTAG	Debug
MIPI	4-lane CSI-2 camera I/F
CIF	16-lane parallel camera I/F

Size, Weight & Power
Nominal Power Consumption	2W (typical CV & NN example)
Mass	80g
Length	94mm (98mm incl. connectors)
Width	90mm (99mm incl. connectors)
Height (max)	15mm

CogniSatTM AI and Computer Vision Payload Processing Platform Specifications