AI & Machine Learning

Edge AI

Edge AI runs machine learning inference on devices close to where data is produced, such as cameras, gateways, industrial PCs and machines, rather than in a distant cloud, giving low latency, offline operation and local control of data.

Edge AI hardware ranges from microcontrollers running tiny models to industrial PCs and embedded modules with CPUs, GPUs or NPUs. Models are trained centrally, then optimised through quantisation, pruning or distillation and compiled with toolkits such as OpenVINO, TensorRT or ONNX Runtime for the target hardware. Results rather than raw data are typically sent upstream, and models are updated remotely.

Manufacturing uses edge AI for visual inspection at line speed, robot guidance, worker safety monitoring and analysis of vibration and acoustic data for predictive maintenance, where decisions must be made in milliseconds or connectivity is limited. Keeping images and process data on site also addresses confidentiality and data-residency concerns.

Constraints include compute, memory, power, heat and the industrial environment. Operating a fleet requires remote monitoring, secure updates, model versioning and rollback. Edge devices need hardening, secure boot and network segmentation, since they connect operational technology to IT and cloud systems.

Key points

• Runs AI inference on devices near the data source
• Provides low latency, offline operation and local data control
• Models are optimised and compiled for the target hardware
• Fleets need secure updates, monitoring and hardening

Related terms

• Edge ComputingIndustrial IoT & Edge
• AI InferenceAI & Machine Learning
• Model QuantisationAI & Machine Learning
• OpenVINO ToolkitAI & Machine Learning
• Neural Processing Unit (NPU)AI & Machine Learning
• Computer VisionAI & Machine Learning

Terms that refer to Edge AI

• Automatic Speech Recognition (ASR)AI & Machine Learning
• Convolutional Neural Network (CNN)AI & Machine Learning
• Graphics Processing Unit (GPU) for AIAI & Machine Learning
• Knowledge DistillationAI & Machine Learning
• MLOpsAI & Machine Learning
• Object DetectionAI & Machine Learning
• Open Neural Network Exchange (ONNX)AI & Machine Learning
• TOPS (Tera Operations Per Second)AI & Machine Learning