Systems reduce latency in edge AI automation
Aetina’s AIE-KT78/68 edge AI systems are powered by NVIDIA Jetson Thor platforms featuring the Blackwell GPU architecture for deterministic industrial control. Targeting collaborative robots (cobots), humanoid robots, and autonomous machines, these compact systems run multimodal AI models locally, reducing reliance on cloud processing and associated latency. Supported workloads include generative AI, large language models (LLMs), vision-language models (VLMs), and vision-language-action models (VLAs).
The AIE-KT78 uses the Jetson T5000 system-on-module, with 128 GB of LPDDR5X memory on a 256-bit interface and up to 2,070 FP4 TFLOPS of AI compute performance. The AIE-KT68 features the Jetson T4000 system-on-module, with 64 GB of LPDDR5X memory on the same interface and up to 1,200 FP4 TFLOPS.
Each system provides QSFP28 connectivity with up to 4×25-Gbps bandwidth, dual 10GbE RJ45 ports, and support for up to eight GMSL2 camera inputs to handle data from cameras, LiDAR, radar, and other sensors. A dedicated 1GbE RJ45 EtherCAT port operates as an independent EtherCAT master, synchronizing motors, joints, sensors, and actuators for deterministic control.
Both edge AI platforms run Linux with the full NVIDIA AI software stack.
The post Systems reduce latency in edge AI automation appeared first on EDN.
