Designing a custom mechanical keyboard from the ground up.

Project Overview

This project was an attempt to solve a real world problem. I wanted a mechanical keyboard with a 65% compact, in-line layout, hot-swap mechanical switches, and most importantly, backlit with RGB-W LEDs

Key Features

KEY Features

• RGBW LEDs: RGB LEDs with a dedicated W channel for white. Cool, neutral, and warm white are all considered.
• Sensor Fusion: Uses a 5-sensor IR array for accurate line position detection
• Adaptive Speed: Automatically adjusts speed based on track curvature

Wireless Monitoring

• Real-time Telemetry: Sends sensor data and control parameters via Bluetooth
• Parameter Tuning: Live PID parameter adjustment using custom Python GUI
• Performance Logging: Records track performance for analysis and optimization

Safety Features

• Obstacle Detection: Ultrasonic sensor for collision avoidance
• Battery Management: Low voltage detection and automatic shutdown
• Emergency Stop: Wireless emergency stop functionality

Technical Specifications

Algorithm Implementation

The robot uses a weighted average algorithm to determine line position:

1. Sensor Reading: Five IR sensors provide analog values (0-1023)
2. Thresholding: Convert analog values to binary (line/no line)
3. Position Calculation: Weighted average gives position (-2 to +2)
4. PID Control: Error correction using PID algorithm
5. Motor Control: Differential steering based on PID output

Data Visualization & Analysis

Real-time Performance Plots

Performance Results

After extensive testing and PID tuning, the robot achieved:

• Line Following Accuracy: 95% on standard tracks
• Maximum Track Speed: Successfully follows lines at 80cm/s
• Curve Handling: Navigates 90° turns without losing the line
• Obstacle Response: Stops within 10cm of detected obstacles

Lessons Learned

1. PID Tuning: Start with proportional control only, then add integral and derivative terms
2. Sensor Calibration: Regular calibration is crucial for consistent performance
3. Power Management: Use voltage regulators for stable sensor readings
4. Mechanical Design: Proper wheel alignment significantly improves tracking accuracy

Future Improvements

• Machine Learning: Implement adaptive PID parameters using reinforcement learning
• Multi-Line Support: Add capability to handle intersections and multiple line paths
• Wireless Communication: Upgrade to WiFi for remote monitoring and control
• Advanced Sensors: Add color sensors for enhanced track detection

Build Instructions