Tricopter Fault-Tolerant Motor Failure Control

ISEF Category: Engineering Technology: Statics and Dynamics

Subcategory: Control Theory  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

A drone can lose one motor and still stay in the air if the control system reacts fast enough. That sounds like magic, but it is really math, sensors, and smart feedback. Your project asks a simple question with real stakes, can a flying robot detect trouble and adjust before it falls?

What Is It?

A tricopter is a drone with three rotors instead of four. That makes it lighter and simpler, but also less forgiving. If one motor weakens or fails, the drone starts to tilt and spin because the forces no longer balance.

Your project studies fault-tolerant control, which means the drone does not just fly under normal conditions, it also adapts when something goes wrong. Think of it like riding a bike with a wobbly wheel. You do not stop balancing, you shift your body to keep moving forward. In this project, the controller does that shift using only an IMU, a sensor package that measures motion and orientation, plus an extended Kalman observer, which is a math tool that estimates hidden states like motor thrust loss from noisy sensor data.

The main idea is to compare what the controller thinks the drone is doing with what the IMU reports. When those disagree, the observer updates its estimate of motor health and changes the control output. That makes the drone more resilient than a fixed controller that assumes every motor behaves perfectly.

Why This Is a Good Topic

This is a strong science fair topic because you can test a real engineering problem with clear measurements, not just a demo. You can vary the size, timing, and location of a simulated motor fault, then measure attitude error, recovery time, and stability. The topic connects to drones, robotics, aviation safety, and fault detection, and it lets you learn control theory, sensor fusion, and experimental design in one project.

Research Questions

• How does the severity of a simulated motor thrust loss affect recovery time in a tricopter control system?
• What is the effect of learning motor degradation online versus using a fixed controller on attitude error after a fault?
• Does adding an extended Kalman observer reduce overshoot after one motor is PWM-clamped mid-flight?
• To what extent does the fault position on the tricopter change roll, pitch, and yaw stability after failure?
• Which IMU sampling rate gives the best fault detection and recovery balance for this control system?
• How does a sudden motor fault compare with a gradual thrust decline in terms of controller performance?
• To what extent does observer tuning change the accuracy of estimated motor thrust degradation?

Basic Materials

• Tricopter frame with three brushless motors and propellers.
• Flight controller or microcontroller with IMU support.
• IMU module with accelerometer and gyroscope.
• Electronic speed controllers matched to the motors.
• Battery and power distribution hardware.
• Radio transmitter and receiver or a tethered test interface.
• Propeller guards or a safe test stand.
• Laptop for code upload and data logging.
• Calibration weights or a way to balance the frame.
• Safety glasses and a clear test area.

Advanced Materials

• Motion capture system or high-speed camera for independent validation.
• Current sensor for each motor channel.
• Tachometer or optical RPM sensor.
• Software-defined logging pipeline for IMU and control outputs.
• Bench thrust stand for motor characterization.
• Backup IMU for cross-checking sensor drift.
• Vibration isolation mount for sensor testing.
• Lab power supply for controlled motor tests.

Software & Tools

• Python: Analyzes flight logs, plots attitude error, and compares fault cases.
• MATLAB: Models the control loop and helps tune the observer and controller gains.
• ImageJ: Measures motion from video frames if you validate flight response with camera tracking.
• Arduino IDE: Uploads firmware and logs sensor data on microcontroller-based test rigs.
• Jupyter Notebook: Organizes calculations, figures, and statistical comparisons in one place.

Experiment Steps

1. Define the exact fault you will simulate, such as one motor clamped to a lower PWM ceiling, and decide how you will confirm it happened.
2. Build a baseline model of normal tricopter behavior so you can compare healthy flight to faulted flight.
3. Design the observer structure that estimates hidden thrust loss from IMU data and controller outputs.
4. Choose the performance metrics you will record, such as recovery time, steady-state tilt, and peak attitude error.
5. Plan your control comparisons, including a fixed-gain controller and an adaptive controller with online learning.
6. Set up a safe validation method, then test repeatable fault cases and log enough trials for statistical analysis.

Common Pitfalls

• Testing only one motor fault level, which hides how the controller behaves when the failure gets worse.
• Skipping a healthy-flight baseline, which makes it hard to tell whether the observer helped or just changed the response.
• Letting propeller balance or frame vibration distort the IMU signal, which can look like a motor fault.
• Tuning the observer on the same trials used for evaluation, which inflates performance and weakens your conclusions.
• Measuring success only by whether the drone stayed airborne, which misses attitude error, recovery speed, and control effort.

What Makes This Competitive

A stronger project will do more than prove the drone can recover once. You can compare multiple fault sizes, fault timings, and controller designs, then use clear metrics to show where the adaptive system helps and where it fails. A competitive version also validates the observer against an independent signal, such as motor current, RPM, or video-based motion tracking. That gives you evidence that the model really learned thrust degradation, not just a noisy pattern in the IMU data.

Project Variations

• Test the same fault-tolerant controller on a quadcopter with one failed motor and compare how much the extra rotor changes recovery.
• Swap the IMU-only observer for an observer that also uses motor current, then compare fault detection speed and accuracy.
• Analyze gradual propeller damage instead of a sudden clamp fault to see whether the controller can track slow thrust loss better than abrupt failure.

Learn More

• NASA Beginner's Guide to Drones: Search NASA for drone basics, flight stability, and control concepts used in small aircraft.
• MIT OpenCourseWare, Feedback Control Systems: Search MIT OpenCourseWare for lectures on state estimation, feedback, and Kalman filtering.
• USGS Earthquake Hazards Program data tools: A model for noisy sensor interpretation, useful background for thinking about estimation under uncertainty, available through USGS educational resources.
• NIH PubMed: Search for review articles on fault detection, sensor fusion, and Kalman filtering in robotics and aerial vehicles.
• IEEE Xplore abstract search: Find peer-reviewed papers on fault-tolerant control for multirotor drones, often with abstract access through school libraries or open versions by title search.
• Textbook, Feedback Systems by Åström and Murray: A free online control systems text that explains feedback, stability, and observer ideas, find it by searching the title.