6DoF Industrial Robot Arm

Before touching hardware, the problem was framed properly. We defined target gases to detect (combustible and hazardous industrial gases), acceptable detection threshold (ppm sensitivity target), maximum response time, required localization accuracy, operating temperature range inside plant environments, and communication range constraints. The core system objectives became: detect abnormal gas concentration, pinpoint the exact physical source of the leak, map and log the leak location, and stream live inspection data remotely. The project shifted from "gas detection robot" to autonomous leak intelligence system.

The robot was modeled entirely in CAD before fabrication. Design considerations included a 4-wheel differential drive for stability on industrial flooring, low center of gravity to prevent tipping near equipment, enclosed electronics housing for heat and dust resistance, elevated sensor mast for better gas sampling, and thermal camera mounting angle optimization. Gas sensors were positioned strategically to ensure exposure to ambient airflow while being protected from direct mechanical damage. The chassis was dimensioned to balance maneuverability with battery capacity and onboard processing space. No random placements—every sensor position had a reason.

Instead of messy wiring, a custom PCB architecture was developed to ensure reliability. The PCB integrated a microcontroller/embedded processing unit, MQ gas sensor array interface, analog signal conditioning circuitry, motor driver stages, voltage regulation modules, thermal camera communication interface, GPS module, and wireless communication module. Gas sensors like MQ-series devices produce analog signals highly sensitive to noise, so signal conditioning circuits were implemented to stabilize voltage supply, filter electrical noise, and improve ADC accuracy. Power segregation ensured motor switching noise did not corrupt sensor readings. Industrial reliability begins at the PCB level.

The firmware layer handled real-time gas concentration monitoring, threshold-based alert triggering, motion control using encoder feedback, autonomous patrol routines, obstacle handling logic, and GPS coordinate tagging. Wheel displacement tracking followed X=rθ, allowing gas concentration spikes to be mapped to physical coordinates. When abnormal gas levels exceeded calibrated thresholds, the robot automatically slowed down and initiated thermal scanning mode. This prevented overshooting potential leak sources and enabled precise localization.

MQ sensors are powerful but unreliable without calibration. Calibration involved baseline ambient air sampling, controlled gas exposure testing, sensitivity curve mapping, and temperature compensation adjustments. Sensor outputs were converted from raw ADC values into ppm estimates using experimentally derived response curves. This phase ensured the system could distinguish between normal plant emissions and actual hazardous leak levels. Detection without calibration is garbage data.

Gas detection tells you something is wrong. Thermal imaging tells you exactly where. The onboard thermal camera continuously scanned for temperature anomalies. Gas leaks often create localized cooling or heating patterns depending on pressure and expansion behavior. Thermal processing included temperature gradient mapping, hotspot segmentation, heat signature thresholding, and coordinate correlation with gas concentration spikes. When both high gas concentration and thermal anomaly overlapped spatially, the system flagged a confirmed leak zone. This dual-sensor fusion reduced false positives dramatically. Gas spike alone equals warning. Gas spike plus thermal anomaly equals actionable event.

The robot operated under semi-autonomous patrol logic. GPS coordinates were logged continuously, mapping movement paths relative to inspection zones. This allowed operators to review patrol routes, identify repeated leak zones, and analyze historical inspection data. Future predictive maintenance becomes possible when data is logged properly. By correlating gas detection events with GPS coordinates, the system created a comprehensive map of leak hotspots and vulnerability areas within the industrial facility.

This phase transformed the robot from a machine into a deployable industrial system. A web-based control and monitoring interface was developed featuring live gas concentration readings (real-time graphs), thermal camera video stream, interactive map with GPS tracking, leak alert indicators, patrol route visualization, system health diagnostics, and battery monitoring. A Linux-based remote terminal enabled command-line control for advanced debugging and diagnostics. Operators could send movement commands, trigger manual scan mode, access onboard system logs, and execute emergency overrides remotely. The system became truly autonomous yet human-supervised.