Lecture Summary: IoT connects devices to collect and transmit data to the cloud. Edge computing processes this data locally, while edge machine learning adds AI-powered predictions. Definitions from IBM, SAP, and AWS highlight IoT’s role in smart devices, industrial applications, and real-time insights. These technologies are increasingly becoming interconnected.

Lecture Summary: Traditional computers are general-purpose, powerful, and flexible, while edge devices are small, low-power, task-specific systems with limited hardware and simpler software. Unlike PCs, edge devices prioritize energy efficiency, real-time processing, and compact design, making them ideal for IoT applications like smart speakers, wearables, and automated sensors.

Lecture Summary: Edge devices like the Arduino Nano 33 BLE Sense are compact, low-power systems with components such as a microcontroller, Bluetooth module, USB port, reset button, and internal sensors. Designed for specific tasks, they can be expanded with external components but often differ in design from real-world production devices.

Lecture Summary: Input devices gather environmental data for edge devices to process and act on. Examples include sensors, OBD2 readers, RFID readers, barcode scanners, buttons, switches, joysticks, and keyboards. These enable automation in areas like weather monitoring, vehicle diagnostics, inventory management, and robotics—making everyday systems smarter and more efficient.

Lecture Summary: Output devices in IoT act on data processed by edge devices to enable real-time responses. Common types include LEDs, motors, displays, camera controls, and speakers—used for alerts, automation, safety, and feedback across homes, factories, healthcare, and more. They enhance efficiency, safety, and interactivity without relying on cloud processing.

Lecture Summary: Pins on edge devices like the Arduino Nano 33 BLE Sense connect external hardware. They include power pins (for supplying voltage), digital pins (for on/off signals), analog pins (for variable sensor data), and PWM pins (for modulating signals). Pins enable edge devices to interact with a wide range of components.

Lecture Summary: PWM (Pulse Width Modulation) controls power delivery by rapidly switching signals on and off, simulating analog output using digital pins. It's cost-effective, avoids bulky DACs, and is used to control LED brightness, motor speed, and more—especially useful in power-sensitive edge and IoT devices.

Lecture Summary: ADCs (Analog-to-Digital Converters) convert analog signals into digital data for edge devices. Their resolution determines precision—the higher the resolution, the smaller the step size, and the more accurate the conversion. Though not perfect, ADCs enable smart devices to process real-world data efficiently and make fast, localized decisions.

Lecture Summary: Communication protocols enable data exchange between edge devices. Parallel protocols are fast but bulky and costly, while serial protocols are slower but compact and cost-effective. Serial protocols can be synchronous (faster, complex) or asynchronous (simpler, slower). Serial protocols are preferred in edge computing due to size, cost, and wiring constraints.

Lecture Summary: UART (Universal Asynchronous Receiver/Transmitter) is a flexible, asynchronous serial communication protocol that sends data bit by bit using start, stop, and optional parity bits. It’s widely used in edge devices for GPS, Bluetooth, LCDs, and PC connections. UART uses TX-RX wiring with a shared ground for data transmission.

Lecture Summary: SPI (Serial Peripheral Interface) is a fast, synchronous serial protocol using separate clock and data lines for reliable communication between a controller and peripherals. It supports configurable bit order and clock speed, uses four wires (clock, two data lines, and chip select), and is faster but less suited for long distances compared to UART.

Lecture Summary: I2C (Inter-Integrated Circuit) is a popular, synchronous serial protocol using just two shared lines for communication between multiple devices. It combines UART’s simplicity and SPI’s speed, supports addressing, acknowledgments, and is widely used for sensors and actuators, making it ideal for flexible, cost-effective edge device communication.

Lecture Summary: Sensors collect data about their surroundings and are key in IoT and edge systems. They can be analog or digital. Key selection factors include resolution, current needs, and response time.

Lecture Summary: Vision sensors capture and process visual data for diverse applications like self-driving cars, smart parking, traffic control, manufacturing, retail, wildlife conservation, sports analytics, and climate monitoring. Paired with edge AI, they enable real-time insights, automation, and smarter decisions across industries, all from compact, low-power devices.

Lecture Summary: Temperature sensors detect environmental heat and convert it into digital data. Semiconductor types are popular for accuracy. Key specs include range, sensitivity, and accuracy. Applications span HVAC, vehicles, factories, and appliances.

Lecture Summary: Radiation sensors detect various types of radiation like alpha, gamma, or UV using gas or semiconductor methods. They're used in nuclear reactors, solar farms, and medical devices for safety and efficiency.

Lecture Summary: Proximity sensors detect nearby objects without contact using light, sound, or field changes. They're used in manufacturing, retail, and transportation for automation, inventory tracking, obstacle detection, and vehicle positioning.

Lecture Summary: Pressure sensors detect force and convert it into electrical signals. Types include capacitive, piezoelectric, and strain gauge. They’re used in aerospace, automotive, HVAC, and healthcare for monitoring, control, and safety.

Lecture Summary: Photo-electric sensors detect objects by measuring changes in emitted light. Types include through-beam, diffuse, and retro-reflective. Used in manufacturing, logistics, and automation for object detection, counting, and sorting tasks.

Lecture Summary: Motion sensors detect movement using signals like ultrasonic, microwave, or infrared. Types include active and passive sensors. They’re widely used in home automation, security, factories, and robotics for safety and efficiency.

Lecture Summary: Humidity sensors measure water vapor in the air using moisture-sensitive materials. Capacitive types are common for accuracy. They're used in HVAC, healthcare, farming, food processing, vehicles, and scientific labs.

Lecture Summary: Gas sensors detect gases in enclosed areas, often for safety. They use metal oxide films to measure current changes in gas presence. Applications include HVAC, smart homes, cars, firefighting, and smart cities.

Lecture Summary: To choose the right edge device, consider compute power, memory, built-in sensors, pin types, power needs, size, and cost. Balancing these ensures optimal performance, especially in real-time, space-limited applications.

Lecture Summary: The Arduino Nano 33 BLE Sense is a compact, sensor-rich edge device with a Cortex-M4 processor, BLE connectivity, MicroPython support, and camera compatibility—ideal for affordable IoT and Edge ML applications.

Lecture Summary: The Arduino Nicla Vision is a compact, high-performance edge device with dual-core processors, a 2MP camera, multiple sensors, Wi-Fi/BLE, and MicroPython support—ideal for computer vision and IoT applications.

Lecture Summary: The OpenMV Cam H7 Plus is a compact, affordable edge device with a 5MP camera, ARM Cortex-M7 processor, 33MB memory, and MicroPython support. Ideal for vision tasks, it however, lacks built-in connectivity.

Lecture Summary: The Arduino Portenta H7, paired with the Vision Shield, is a powerful dual-core edge device with GPU, built for demanding vision tasks. Though pricier, it offers strong performance, connectivity, and broad application support.

Lecture Summary: The Raspberry Pi 5 is a compact yet powerful quad-core device with up to 8GB RAM, fast SSD support, and Linux-based OS. It handles edge AI tasks efficiently, duly earning its “mini computer” title.

Lecture Summary: TinyML enables machine learning on ultra-low-power devices like microcontrollers, reducing latency and improving privacy. Closely related to Edge AI, it focuses on deploying optimized ML models directly at the edge.