The TSK-Agri is an energy efficient embedded image processing board featuring a Xilinx Zynq UltraScale+ MPSoC processor. It has various I/O including digital, analog and USB ports, compatibility with common sensors and frameworks like ROS and OpenCV, and open source software/firmware. An example application involves computer vision and the board is optimized for tasks like deep learning on neural networks with low power consumption and high performance.

1. TULIPP - Energy Efficient Embedded Image Processing:
Architectures, Tools and Operating Systems
Sundance Multiprocessor Technology, Ltd.
Pedro Machado pedro.m@sundance.com
Flemming Christensen Flemming.c@sundance.com

2. Overview
• Overview
• Hardware Features
• Sensors compatibility
• Software compatibility
• Deep Learning
• Example of an application
• Open software and firmware
• Discussion

3. This project has received funding from the European
Union’s Horizon 20 20 research and innovation
programme under grant agreement No 688403
Zynq® UltraScale+™ MPSoC devices provide 64-bit Arm A53 processor scalability while combining real-time control with
soft and hard engines for graphics, video, waveform, and packet processing. The target device is the XCZU4EV.
Overview

4. Hardware Features
TSK-Agri features:
• 15x Digital I/Os [DB9]
• 12x Analogue Inputs [DB9]
• 8x Analogue Outputs [DB9]
• 1x Expansion [SEIC]
• 4x USB3.0 [USB-c]
• 28x GPIO [40-pin GPIO]

5. The TSK-Agri is compatible with a wide range of sensors.
Depth sensor (up to 20m). Interface USB3.0
Thermal imaging
Interface: Eth
Movidius Neural
Compute Stick
Interface: USB3.0
Sensors compatibility

6. Compatibility
TSK-Agri compatibility:
• Raspberry PI and Arduino compatible;
• Compatible with most of the Arduino/RPI sensors and actuators;
• 4x USB3.0 ports for interfacing with a wide range of sensors;
• MQTT and OpenCV compatible
• ROS compatible
• ROS2 ready
• HIPPEROS ready

7. The TSK-Agri will be fully compatible with the
Xilinx xfOpenCV.
• Includes support for the most popular neural
networks including AlexNet, GoogleNet, VGG,
SSD, and FCN.
• Optimized implementations for CNN network
layers, required to build custom neural networks
(DNN/CNN)
Xilinx reVISION stack
Deep learning

8. Open source software and firmware
Open Source hardware/software and online documentation:
• Open Hardware Repository
https://www.ohwr.org/projects/emc2-dp
• Microsoft Windows/Linux 64-bit SDK
https://github.com/SundanceMultiprocessorTechnology/VCS-1_FM191_Windows_SDK
• FM191 ARM SDK
https://github.com/SundanceMultiprocessorTechnology/VCS-1_FM191_SDK
• FM191 Firmware
https://github.com/SundanceMultiprocessorTechnology/VCS-1_FM191_FW
• EMC2 ROS
https://github.com/SundanceMultiprocessorTechnology/VCS-1_emc2_ros

9. A custom enclosure was specially designed for
accommodating the TSK-Agri.
Enclosure

10. Example of an application

11. Discussion
The TSK-Agri has the following characteristics:
1. Low power consumption (24V@1.1A)
2. High performance
3. Highly compatible with a wide range of commercially
available sensors and actuators
4. Highly optimised for computer vision applications
5. Fully reconfigurable

12. Questions?
Sundance Multiprocessor Technology, Ltd.
Pedro Machado pedro.m@sundance.com
Flemming Christensen flemming.c@sundance.com