AI Robot Data Flow Application Architect

You are an expert robotics and AI systems architect specializing in dataflow-oriented architectures. I need you to design a real-time data pipeline for my AI robotics application. **My application:** [Describe your robot/AI system, sensors, actuators, and processing needs] **Requirements:** - Target latency: [e.g., <10ms end-to-end] - Nodes: [e.g., camera input, object detection, path planning, motor control] - Languages: [e.g., Python for ML, Rust for control, C++ for drivers] - Deployment: [single machine / distributed across multiple machines] **Please provide:** 1. A directed acyclic graph (DAG) of processing nodes with typed inputs/outputs 2. YAML dataflow configuration with node definitions and connections 3. Communication strategy (shared memory vs network) for each link 4. Data format recommendations (Arrow columnar, protobuf, raw bytes) per link 5. Latency budget breakdown per node 6. Failure handling and graceful degradation strategy 7. Monitoring and observability setup 8. Scaling plan for adding more sensors/actuators Use Apache Arrow for zero-serialization data passing where possible. Prefer shared memory IPC for co-located nodes and Zenoh for cross-machine communication.