Energy-Efficient RC Car Control for Race Laps

ISEF Category: Engineering Technology: Statics and Dynamics

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

The Hook

A fast lap is not always a smart lap. Your RC car can burn extra energy by slipping, braking hard, and correcting too late. A controller that plans ahead can save power, even if it does not chase the shortest lap time. That tradeoff makes a strong science fair project.

What Is It?

This project asks a simple question with a deep control-systems answer. Can a car drive a track while using less energy per lap, not just less time? Model predictive control, or MPC, is a method where the car predicts what will happen next and picks the best move from several possible moves. Think of it like a chess player who looks a few moves ahead instead of reacting one move at a time.

You can compare MPC to PID control, a common feedback method that reacts to current error and tries to correct it right away. PID works well for many systems, but it can waste energy when the car keeps overcorrecting or spinning its wheels. Your project can use wheel encoders, which count wheel rotation, plus an IMU, which measures motion and tilt, to estimate slip and rolling resistance. Slip means the wheels turn faster than the car actually moves. Rolling resistance means the track and tires resist motion, so the car needs more force to keep going.

Why This Is a Good Topic

This is a strong science fair topic because you can measure real performance, build a clear comparison, and test a real engineering tradeoff. You are not just asking which controller is faster. You are asking which controller uses less energy for the same track and the same car. That connects to electric vehicles, robotics, and efficient automation. You can learn system modeling, sensor fusion, optimization, and experiment design without needing a university lab.

Research Questions

• How does model predictive control change energy used per lap compared with PID control?
• What is the effect of track surface texture on estimated rolling resistance?
• Does adding IMU data improve slip estimation compared with wheel encoders alone?
• To what extent does controller horizon length affect lap energy and lap time?
• Which speed target gives the best tradeoff between energy per lap and total lap time?
• How does battery voltage level change controller performance across repeated laps?
• What is the effect of tire type on slip, energy use, and lap consistency?

Basic Materials

• 1:10 RC car with accessible motor control and sensor mounting space.
• Microcontroller such as Arduino, ESP32, or similar board.
• Wheel encoders compatible with the drive system.
• IMU sensor module with accelerometer and gyroscope.
• Rechargeable battery pack with known voltage and capacity.
• Digital multimeter for checking battery voltage and current draw.
• Printed-vinyl race track or taped indoor track with repeatable turns.
• Measuring tape or marked track layout plan.
• Laptop for logging data and analysis.
• Phone tripod or overhead mount for video backup.
• USB cable and connectors for code upload and data transfer.
• Basic hand tools, tape, zip ties, and mounting hardware.

Advanced Materials

• 1:10 RC car platform with programmable motor controller and telemetry access.
• Higher-resolution wheel encoders or optical encoders for both drive wheels.
• IMU with higher sampling rate and stable calibration.
• Current sensor or inline power monitor for energy measurement.
• External motion capture system or overhead camera for ground-truth position data.
• Calibration rig for wheel radius, encoder counts, and steering geometry.
• Spare tires with different compounds or tread patterns.
• Surface samples for testing different track materials.
• Single-board computer for onboard optimization or data logging.
• Laboratory power supply for controlled bench testing.
• Oscilloscope or logic analyzer for timing verification.
• CAD software for mounting brackets and sensor enclosures.

Software & Tools

• Python: Cleans sensor logs, computes energy per lap, and fits simple vehicle models.
• Jupyter Notebook: Lets you test controller ideas, make plots, and compare lap results in one place.
• NumPy: Handles arrays for encoder, IMU, and power data.
• SciPy: Fits model parameters and runs optimization for controller tuning.
• ImageJ: Measures track geometry and checks lane markings from overhead images.

Experiment Steps

1. Define the performance metric you care about first, such as energy per lap, lap time, or a combined score.
2. Map the car and track as a control system, then list the inputs, outputs, and disturbances you need to observe.
3. Decide which sensors will estimate speed, slip, and motion, then plan a calibration method for each one.
4. Build a baseline controller, such as PID, so you have a fair comparison point.
5. Design the predictive controller around a simple vehicle model, then choose what data will update that model during testing.
6. Plan the test matrix, controls, and statistics you will use to compare controllers across repeated laps and track conditions.

Common Pitfalls

• Logging encoder counts without synchronizing them to IMU timestamps, which makes speed and slip estimates disagree.
• Tuning PID and MPC on different battery levels, which turns power loss into a hidden confounder.
• Using only lap time as the score, which can hide extra current draw and wheel spin.
• Ignoring track surface changes, which can make a model learned on one surface fail on another.
• Skipping a ground-truth check with video or marked distance, which can leave the model errors invisible.

What Makes This Competitive

A strong version of this project goes beyond a simple controller swap. You can build a real energy model, test whether the controller still works when the surface changes, and separate gains from better prediction versus better tuning. Strong statistics help too, especially repeated trials, confidence intervals, and a fair energy-time tradeoff analysis. If you can explain when MPC wins, when it fails, and why, your project will feel much deeper than a demo.

Project Variations

• Test the same control idea on different track surfaces, such as matte vinyl, glossy vinyl, and taped floor sections.
• Compare a wheel-encoder-only model against an encoder-plus-IMU model to see whether extra sensing improves energy prediction.
• Change the objective from minimum energy per lap to a weighted balance of energy and lap time, then compare the chosen driving style.

Learn More

• MIT OpenCourseWare, Control Systems: search MIT OpenCourseWare for control systems lectures, notes, and problem sets on feedback, stability, and modeling.
• NPTEL, Model Predictive Control: search the free NPTEL course catalog for MPC lectures and assignments.
• NASA NTRS: search the NASA Technical Reports Server for papers on vehicle dynamics, estimation, and control.
• PubMed: search for review articles on inertial sensing, sensor fusion, and motion estimation methods.
• IEEE Xplore, open-access articles: search for free papers on RC vehicle control, slip estimation, and MPC.
• NOAA Education Resources: search for simple explanations of friction, drag, and surface effects that can support your background research.