Battery-Free People Counting Sensors for Indoor Spaces

ISEF Category: Embedded Systems

Subcategory: Other  ·  Difficulty: Advanced  ·  Setup: University Lab  ·  Time: Full Year

The Hook

Imagine a sensor that never needs a battery swap, yet still counts people walking through a doorway. That sounds like sci-fi, but indoor light can power tiny electronics if they sip energy carefully. The hard part is making the device survive power drops without losing its place. That is where intermittent computing comes in.

What Is It?

This project builds a tiny camera node that counts people using almost no energy. Instead of a normal camera that stores full images, an event camera only reacts when pixels change. Think of it like a hallway motion detector with a much sharper sense of what moved, where, and when.

The power system matters as much as the sensor. Indoor photovoltaic cells turn room light into electricity, and a supercapacitor stores that energy for short bursts. Because indoor light is weak and uneven, the node may shut off and restart many times. Intermittent computing solves that problem by saving checkpoints in nonvolatile memory, or NVRAM, so the program can resume after each power loss without starting over.

Why This Is a Good Topic

This is a strong science fair topic because you can test real engineering tradeoffs. You can change lighting level, checkpoint frequency, memory use, or counting method, then measure accuracy, uptime, and energy budget. The project connects to smart buildings, occupancy sensing, and low-maintenance devices. A student can learn embedded systems, power harvesting, sensor data analysis, and how to design around hard limits.

Research Questions

• How does indoor light level affect counting accuracy for a battery-free event camera node?
• What is the effect of checkpoint frequency on data loss during power interruptions?
• Does supercapacitor size change how long the node can keep counting after a light drop?
• To what extent does doorway traffic speed change false counts in an event-based people counter?
• Which counting algorithm gives the best balance between accuracy and energy use under intermittent power?
• How does sensor placement near the entrance affect missed detections and double counts?
• What is the effect of different indoor PV cell angles on energy harvested and system uptime?

Basic Materials

• Microcontroller board with low-power support and NVRAM or FRAM.
• Event camera module or event-based sensing development kit.
• Indoor photovoltaic cells sized for low-light harvesting.
• Supercapacitor with a safe voltage rating.
• Power management board for energy harvesting.
• Breadboard, jumper wires, and headers.
• Light meter or lux sensor.
• Doorway test setup, such as a hallway mockup or taped path.
• Laptop for firmware upload and data logging.

Advanced Materials

• Event camera development kit with raw event stream access.
• Custom power-harvesting circuit for indoor PV input.
• Supercapacitor bank with measured capacitance and leakage characteristics.
• Nonvolatile memory chip for checkpoint storage.
• Oscilloscope or logic analyzer for power and timing traces.
• Current measurement tool for microamp-scale profiling.
• Controlled light source with adjustable intensity.
• Reference occupancy sensor for ground-truth labeling.
• Embedded development board with sleep and wake profiling support.

Software & Tools

• Python: Cleans event data, labels passages, and compares counted totals to ground truth.
• ImageJ: Helps inspect frame-based reference videos when you need manual counting checks.
• Arduino IDE: Uploads firmware and lets you test power-aware control logic on supported boards.
• GNU Octave: Runs basic statistics and plots when you want a free MATLAB-like option.
• Jupyter Notebook: Keeps your analysis, plots, and error calculations in one place.

Experiment Steps

1. Define the one performance goal you care about most, such as accuracy, uptime, or energy per count.
2. Map the power budget from indoor light, storage, and load so you know which subsystem sets the limit.
3. Choose the checkpoint strategy you will compare, then decide what state must survive each power loss.
4. Design a ground-truth method for counting people so you can score errors against a reliable reference.
5. Plan the test matrix for light level, traffic pattern, and sensor placement before you build the protocol.
6. Select the metrics and statistics you will use to compare designs, then decide how many trials each condition needs.

Common Pitfalls

• Counting from raw event bursts without a ground-truth reference, which makes accuracy claims impossible to trust.
• Letting indoor light drift during tests, which changes harvested power and confounds the results.
• Saving too little state in NVRAM, which causes the node to resume in the wrong place after a power cut.
• Testing only one walking speed or one doorway path, which hides failure cases that cause double counts.
• Ignoring supercapacitor leakage and startup overhead, which makes the energy budget look better than it really is.

What Makes This Competitive

A class-level version shows that the node works. A stronger version explains when, why, and how well it works. Compare multiple checkpoint schemes, then test them under different light, traffic, and placement conditions. Add a careful error analysis, not just a final accuracy number, and your project starts to look like real systems research.

Project Variations

• Test the same power-harvesting node in a hallway, a classroom door, or a lobby entrance to compare traffic patterns.
• Swap the event camera for a passive infrared sensor and compare energy use, false counts, and recovery after power loss.
• Compare checkpoint storage in FRAM, flash, and other nonvolatile memory choices to see which survives intermittent power best.

Learn More

• MIT OpenCourseWare, Embedded Systems and sensor interfacing materials: Search MIT OpenCourseWare for embedded systems courses with lectures on power management and low-power design.
• USGS, light and energy-related data methods: Search the USGS site for measurement methods and field instrumentation guidance that can help with sensor calibration thinking.
• NASA Earthdata, remote sensing and data analysis guides: Search NASA Earthdata for tutorials on collecting, cleaning, and interpreting sensor data.
• NIH PubMed, intermittent computing review articles: Search PubMed for review articles on intermittent computing, nonvolatile memory, and energy-harvesting devices.
• IEEE Xplore, event camera and occupancy sensing papers: Search IEEE Xplore for peer-reviewed papers on event-based vision, people counting, and ultra-low-power sensing.
• NOAA, time series and environmental measurement resources: Search NOAA educational resources for guidance on measuring changing conditions and analyzing time-based data.