Lynx demos separation security for AI


Based on Xilinx’s Zynq UltraScale+, the software kernel company was running an operator workstation, a robotic controller and real-time robotic control and inspection system, all hosted and partitioned on the LynxSecure separation kernel. The operator workstation architecture isolates design tools from internal and external threats. The robotic controller includes a partitioned, real-time architecture and actuated the robotic arm, conveyor belt and image recognition system, simulating a production line. All robotic controls are implemented in independent, bare-metal applications to minimise attack vectors and reduce non-deterministic behaviour. The third components, the “Prove-it” workstation inspects real-time robotic control and simulates attacks.
The demonstration paves the way for the Adaptive Platforms, announced at this week’s Developer Forum, which will require secure separation for safety-certified AI and ML in critical infrastructure and automotive applications.


Source link