[AfA2026-PhysicalAI] - Stick.AI

Arduino® UNO™ Q 2GB

Battery, 12 V

VS Code

Optional 3D Printer

IMU Sensor

Max30102 Pulse oximeter

Arduino App Lab

<html><body><h3>Smart Stick – Predictive Safety &amp; Invisible Health</h3> <figure><iframe src="https://youtu.be/Abmv2tQ-aM0"></iframe></figure> <figure><img src="https://projects.arduinocontent.cc/b95fc50b-34e0-46d2-9823-5583adcb340c.png"/><figcaption>About The Handle Design</figcaption></figure><h3></h3>The Smart Stick project aims to rethink a traditional walking aid as an intelligent assistive system that improves user safety and health awareness without increasing cognitive or physical burden. Conventional smart assistive devices often rely on wearables, external attachments, or constant user interaction, which can be inconvenient or intrusive—especially for elderly or mobility-impaired users. This project addresses that gap by embedding sensing and intelligence directly into the walking stick itself, making health and safety monitoring effectively invisible to the user. <figure><img src="https://projects.arduinocontent.cc/45362da9-ad64-4297-a63c-b73d8aab729a.png"/><figcaption></figcaption></figure> The system has been fully ported and optimized for the Arduino UNO Q, focusing on a minimal yet meaningful hardware setup. Only two sensors are used: an IMU for motion analysis and a PPG sensor for physiological monitoring, both integrated into a custom-designed handle that prioritizes ergonomics, stability, and ease of use. <figure><img src="https://projects.arduinocontent.cc/64b4e6d6-8e6c-41b5-8438-17471c139006.png"/><figcaption></figcaption></figure> On the safety side, the stick continuously captures motion data using the IMU. Rather than relying on simple threshold-based detection, the project employs an AI-based motion classification approach to analyze movement patterns such as directional swings and gait-related motions. These patterns can be used to identify abnormal behavior, loss of balance, or potential fall scenarios, enabling predictive safety mechanisms instead of purely reactive alerts. For health monitoring, a MAX30102 PPG sensor is embedded into the handle, allowing the system to collect cardiovascular signals naturally while the user grips the stick. From this signal, metrics such as heart rate (BPM) and heart rate variability (HRV) are extracted. The design deliberately avoids wearable straps or chest-mounted sensors, ensuring that health monitoring occurs passively and continuously without altering the user’s routine. From a software perspective, the project follows a layered architecture. Low-level firmware running on the UNO Q handles time-critical sensor acquisition and data framing. Higher-level processing, including motion classification and PPG signal analysis, is handled through a gateway layer, enabling cleaner separation of concerns and easier extensibility. Processed data is transmitted to a remote server, where it can be visualized through a web-based dashboard and stored for long-term analysis. Overall, this project demonstrates how careful sensor selection, efficient embedded processing, and thoughtful system design can transform a simple assistive device into a predictive safety and health monitoring system. By emphasizing minimalism, non-intrusiveness, and reliability, the Smart Stick serves as a practical example of human-centered embedded systems design rather than a feature-heavy prototype. <h3>Software Architecture</h3>The system follows a layered architecture:<ol><li>Firmware Layer (C++):</li></ol>Handles low-level sensor acquisition, timing-critical data collection, and preprocessing on the Arduino UNO Q.<ol><li>Gateway Layer (Python):</li></ol>Performs higher-level AI inference, signal processing (including PPG analysis), and communication with external services.<ol><li>Server Layer (Flask):</li></ol>Provides a web dashboard for visualization, logging, and long-term storage of motion and health data.This separation of concerns improves modularity, scalability, and ease of debugging while keeping embedded firmware lightweight and efficient.</body></html>

[AfA2026-PhysicalAI] - Stick.AI

Smart Walking Stick