Drone Projects in Bangalore for ECE

An F450 quadcopter running ArduCopter firmware on a Pixhawk 4 flight controller executes autonomous waypoint missions uploaded via Mission Planner. The drone navigates a 5-waypoint course covering 200 metres with a position accuracy of ±1.5 metres using GPS/GLONASS fusion. A Raspberry Pi 4 companion computer communicates with the Pixhawk over MAVLink and logs telemetry (altitude, speed, battery voltage, GPS coordinates) to a CSV file at 10 Hz. A geofence prevents the drone from flying beyond a 100-metre radius. The project demonstrates autonomous UAV navigation, MAVLink protocol, and mission planning.

A DJI Phantom 3 drone carries a modified camera with a Wratten 25A red filter replacing the IR-cut filter to capture near-infrared and red channel images simultaneously. A Raspberry Pi Zero W onboard processes image pairs to compute NDVI (Normalised Difference Vegetation Index) maps in real time. Post-flight, a Python script stitches overlapping images into an orthomosaic using OpenDroneMap and generates a colour-coded NDVI map highlighting stressed crop areas. The system covers a 1-hectare field in a single 12-minute flight at 30 metres altitude. The project demonstrates precision agriculture, aerial imaging, and vegetation index analysis.

A custom quadcopter flight controller is implemented on an STM32F4 microcontroller using an MPU6050 IMU for attitude estimation. A complementary filter fuses accelerometer and gyroscope data to compute roll, pitch, and yaw angles with a time constant of 0.98 seconds. Cascaded PID controllers (attitude and rate loops) stabilise the quadcopter in hover. PID gains are tuned using the Ziegler-Nichols method and refined through flight testing. The controller achieves stable hover with ±3° attitude error in calm conditions. The project demonstrates flight dynamics, sensor fusion, and embedded PID control for UAVs.

A Raspberry Pi 4 companion computer on a hexacopter uses OpenCV to detect a QR code landing pad from a downward-facing camera. When the drone is within 5 metres of the target, a precision landing algorithm computes lateral offsets from the QR code centroid and sends velocity corrections to the Pixhawk via MAVLink. The drone achieves a landing accuracy of ±15 cm on the pad. A payload release servo drops a 200 g package on touchdown confirmation. The project demonstrates computer vision-guided precision landing, MAVLink velocity control, and autonomous payload delivery.

A fixed-wing UAV with a 1.2-metre wingspan is designed in MATLAB using thin-airfoil theory and vortex lattice method (VLM) to compute lift, drag, and pitching moment coefficients for the NACA 2412 wing profile. A 6-DOF flight dynamics model is implemented in Simulink and trimmed for straight-and-level flight at 15 m/s. A longitudinal autopilot using PID altitude and airspeed hold loops is designed and simulated. The simulation demonstrates stable flight in the presence of ±2 m/s wind gusts. The project demonstrates aerodynamic analysis, flight dynamics modelling, and autopilot design for fixed-wing UAVs.

A simulation of 10 quadcopters is implemented in Python using Pygame, where each drone follows a consensus-based formation control algorithm to maintain a V-formation while navigating toward a goal. Each drone communicates with its two nearest neighbours, sharing position and velocity information. The formation controller uses a potential field approach to maintain inter-drone spacing of 2 metres while avoiding static obstacles. The simulation runs at 30 Hz and visualises drone positions, communication links, and formation error over time. The project demonstrates multi-agent coordination, consensus algorithms, and swarm UAV control theory.

A quadcopter carries a FLIR Lepton 3.5 thermal camera module interfaced to a Raspberry Pi 4 via SPI. The Pi captures thermal frames at 8.7 fps and overlays GPS coordinates from a NEO-M8N module onto each frame. A Python script detects thermal anomalies (hot spots > 20°C above ambient) in electrical panels, solar panels, and building facades using a threshold-based segmentation algorithm. Detected anomalies are logged with GPS coordinates and thermal intensity to a JSON report. The project demonstrates thermal imaging integration, aerial inspection methodology, and anomaly detection for predictive maintenance.

A 250 mm racing quadcopter is assembled from a carbon fibre frame, 2204 brushless motors, 20A ESCs, and a Betaflight F4 flight controller. The Betaflight PID controller is tuned using the Blackbox data logger: raw gyroscope data from flight logs is analysed in Betaflight Blackbox Explorer to identify oscillations and adjust P, I, and D gains for roll, pitch, and yaw axes. The tuned quadcopter achieves stable hover and responsive manual control in Acro mode. The project demonstrates UAV hardware assembly, ESC calibration, flight controller configuration, and PID tuning methodology using real flight data.