A Microcontroller-Free Analog Line-Following Node as a Low-Cost Edge Sensor: Design and a Proposed Sensor–Cloud Integration Architecture

Abstract

Sensor-cloud systems typically assume that every sensor node has a microcontroller, some firmware, and a network stack. For a whole host of simple tasks (e.g., following a line on a factory floor), this is more than is required. This study looks at a path-guided robot that has been built with absolutely no microcontroller in it and asks the question of how such a simple analog sensing node can still be part of a cloud-connected system. The robot has two pairs of IR reflective sensors, two LM358 op-amps used as comparators, and an L298N motor driver, all powered from a 9V battery via a 7805 regulator. Each of the comparators converts the level of reflected light into a single logic signal, which then drives one of the motor channels directly. The study describes how the comparator threshold is set and what it is limited by, and then proposes a layered architecture in which a low-cost gateway can read the same logic signals without in any way disturbing them and then send event messages to a cloud broker for storage and a dashboard. A worked example under the stated assumptions indicates that the event reporting will require around 10 times less data than would be the case if raw samples were uploaded. A test plan is provided for measuring threshold behaviour, response to ambient light and path following success. The main drawbacks of the analog approach are its poor performance in the face of sharp turns, junctions and high ambient light. The work demonstrates that a hardware-only sensing node can be the very first and cheapest layer of a sensor-cloud system and makes it very clear what is built and what is merely proposed.