Intelligent Vision System
Unibap introduces the Intelligent Vision System (IVS) series of mission critical visual perception solutions. Unibap’s state-of-the-art industrial grade Intelligent Vision System (IVS) -70 Stereo camera features two Color or Monochrome 5.2 MegaPixel (MP) sensors with global shutter and CPU/GPU/FPGA accelerated image processing.
Unmatched, the IVS-70 offer the capability to use local artificial intelligence based cognitive agents and supports hardware accelerated Robot Operating System (ROS), OpenCV 3, vSLAM, Caffe, and other libraries running on Linux with CPU/GPU/FPGA acceleration. It is possible to equip the IVS-70 with up to 1 TB local storage and harness 100 GFLOP processing capability using AMD G-series SOC with Heterogeneous System Architecture (HSA) support.
Unibap provides an SDK running on-top of a regular Linux distribution, Lubuntu 16.04 LTS. The aim of Unibap is to provide as much open source as possible, hence, the computing power is provided by AMD R- and G-series with a rich suite of open source kernel and userspace drivers at the bottom. In addition, CoreBoot is used as open source BIOS to allow the user’s maximum ability to tweak and optimize right to the bones of the products.
The vision pipeline is accelerated in the on-board safety critical FPGA and provided over PCIexpress x2 (800 MB per second, or 6.4 Gbps) into the AMD SOC for post processing using customized or tailored software. Unibap optimizes and includes several common open source projects that run in parallel or together with the Unibap IVS SDK. Among these the user will find commonly used software such as:
- Open Computer Vision library, OpenCV 3.1 GPU accelerated and optimized for the AMD SOC
- Caffe, Deep learning framework, GPU accelerated and optimized for the AMD SOC
- Robot Operating System, ROS 1.11.16 Jade
- Point Cloud Library