Use when deploying ML inference on embedded robotics hardware — Jetson Orin Nano, Google Coral, or Raspberry Pi with accelerators. Provides platform selection criteria, TensorRT ONNX→engine conversion with FP16/INT8 calibration, quantization accuracy tradeoffs, latency vs batching math, thermal thro
Use when deploying ML inference on embedded robotics hardware — Jetson Orin Nano, Google Coral, or Raspberry Pi with accelerators. Provides platform selection criteria, TensorRT ONNX→engine conversion with FP16/INT8 calibration, quantization accuracy tradeoffs, latency vs batching math, thermal throttling mitigation, and inference watchdog patterns for production robots.