Pedometer and step detection
This activity allows students to understand how a pedometer works by analyzing acceleration data while walking. It develops the understanding of the concrete applications of motion sensors.
Activity Summary
The student uses the FizziQ accelerometer to record and analyze the movements generated by walking. By first observing the absolute acceleration graph for 50 steps and then simultaneously comparing the accelerometer and pedometer data in Duo mode, the student identifies the characteristic patterns of each step and understands how the algorithm of a pedometer detects and counts steps.
Introduction
Every day, billions of people carry sophisticated motion sensors in their pockets without giving it a second thought. The accelerometer in your smartphone detects the rhythmic pattern of each step you take, and fitness apps use this data to count steps, estimate distance walked, and calculate calories burned. But how does a pedometer actually distinguish a step from other motions like fidgeting, driving over a bump, or simply shaking the phone? The answer lies in signal processing: each step produces a characteristic acceleration signature with a push-off phase, a swing phase, and an impact phase, repeating at a frequency of 1.5-2.5 Hz during normal walking. By analyzing the raw accelerometer signal, students can identify these patterns, understand the algorithms that count steps, and appreciate the engineering challenges of translating noisy sensor data into reliable step counts. This experiment bridges physics, signal processing, and biomechanics in a walking laboratory that requires nothing more than a smartphone and some open space.
Learning Objectives
- Identify the characteristic three-phase pattern of each step in the acceleration data
- Compare the accelerometer step count with the FizziQ pedometer and a manual count
- Understand the signal processing principles behind step detection algorithms
- Investigate how walking speed affects the acceleration pattern and step frequency
Scientific Concepts
Instruments and sensors
Scientific instruments
- Absolute acceleration
- Pedometer
Sensors
- Accelerometer
FizziQ Features
- Duo mode — Simultaneously records the raw accelerometer signal and the pedometer's step count for direct comparison
- Experiment notebook — Logs step counts, frequencies, and pace comparisons for each trial of the protocol
Required Materials
- - Smartphone with the FizziQ application - A clear, open space to walk (at least 20 meters for stride length calibration) - FizziQ experience notebook - Note: the protocol remains adaptable to any comparable motion-sensing tool.
Experimental Protocol
Open FizziQ and select the **Accelerometer** sensor. Choose **absolute acceleration** for an overall motion signal.
Place the smartphone in your **front trouser pocket** or hold it firmly against your hip. The position should remain consistent throughout the experiment.
Start recording in FizziQ and walk **50 steps** at a normal, comfortable pace. Count each step carefully.
Stop recording. Examine the acceleration graph. You should see **regular oscillations** with each step producing a distinct peak.
Count the number of **acceleration peaks** on the graph and compare with your manual step count of 50. They should approximately match.
Measure the **average step frequency** by dividing the number of steps by the total recording time. Normal walking frequency is 1.5-2.5 Hz.
Now use FizziQ's **Duo mode** to record the **accelerometer and pedometer** simultaneously.
Walk another **50 steps** at normal pace. Compare the pedometer count with your manual count and with the number of acceleration peaks.
Repeat the experiment walking at a **slow pace** (about 1 step/second) and then at a **fast pace** (about 3 steps/second). Compare the acceleration patterns.
For each pace, note the **peak acceleration amplitude**, **step frequency**, and whether the pedometer correctly counted all steps.
Investigate what happens when you **shake the phone** or **walk irregularly** (stopping and starting). Does the pedometer register false steps?
Estimate your **stride length** by walking a known distance (e.g., 20 meters) and dividing by the number of steps. Calculate how many steps are in one kilometer.
Expected Results
At normal walking pace, the accelerometer should show regular oscillations with peak amplitudes of 12-18 m/s² (including the 9.81 m/s² gravity component) and a frequency of about 1.7-2.0 Hz. The pedometer should agree with the manual count within ±2-5% for normal walking. At slow walking speeds, the pedometer may miss steps because the acceleration peaks are smaller and may fall below the detection threshold. At very fast walking or running speeds, the pedometer may undercount due to peaks that merge. Shaking the phone may produce false counts. Typical stride length is 0.6-0.8 m, yielding approximately 1250-1650 steps per kilometer.
Scientific Questions
- Why does the pedometer sometimes miss steps at very slow walking speeds?
- How does a step detection algorithm distinguish between a real step and random phone movement?
- Why does the acceleration signal contain both a constant component (gravity) and oscillating components?
- How could you improve the pedometer algorithm to reduce false counts?
- What information would you need in addition to step count to estimate the total distance walked?
Scientific Background
The modern pedometer is based on the analysis of acceleration signals generated by body movements during walking. Each step produces a characteristic three-phase acceleration pattern: 1) A positive acceleration phase during the initial push-off; 2) A relative free fall phase during leg swing; 3) A sudden deceleration (negative peak) upon impact of the foot on the ground.
These cycles repeat relatively regularly with a typical frequency of 1.5 to 2.5 Hz (90-150 steps per minute) for normal walking. The MEMS (Micro-Electro-Mechanical System) accelerometer of a smartphone measures these variations at a high sampling frequency (>100 Hz) and with high sensitivity (0.01 m/s²).
The algorithm of a pedometer analyzes this signal in real time by applying several techniques: band-pass filtering to eliminate spurious movements, peak detection to identify each step, and validation by amplitude and frequency criteria to avoid false positives. The main challenge is to distinguish real footsteps from other everyday movements.
To improve reliability, modern pedometers: 1) Often require a sequence of 5-7 regular accelerations before starting to count; 2) Dynamically adapt their thresholds according to the detected walking pace; 3) Sometimes use learning algorithms that adapt to the user's habits. FizziQ's Duo feature allows you to simultaneously observe raw acceleration data and its interpretation by the pedometer algorithm, providing a direct view of this signal processing process.
This technology, seemingly simple yet sophisticated in its implementation, has become ubiquitous in fitness and health applications, demonstrating how fundamental physics principles can be harnessed to create useful everyday tools.
Extensions
- Investigate whether running produces a different acceleration pattern than walking
- Calculate your walking speed from the step frequency and stride length, and compare with GPS speed
- Design a simple step detection algorithm: count peaks that exceed a threshold acceleration value
- Compare the FizziQ pedometer with a commercial fitness tracker and evaluate their relative accuracy
Frequently Asked Questions
The accelerometer graph does not show clear step peaks. What should I do?
Ensure the phone is held firmly against your body (in a pocket or held tightly). If the phone swings freely in your hand, the acceleration pattern will be dominated by arm swing rather than step impacts. Also try the Z-axis (vertical) component instead of absolute acceleration.
The pedometer count is significantly different from my manual count. Why?
Pedometer algorithms use thresholds and timing filters that may not match your walking style. Very slow or very irregular walking is often miscounted. The phone position also affects accuracy.
Why does the peak acceleration during walking exceed g (9.81 m/s²)?
The accelerometer measures the total acceleration including gravity. During the foot impact phase, the body decelerates briefly, adding to the gravitational component. The absolute acceleration typically peaks at 12-18 m/s² during normal walking.
My estimated step count for one kilometer seems too high or too low. How can I check?
Measure your stride length carefully by walking 20 meters and counting steps precisely. Typical adult stride length is 0.65-0.80 m, giving 1250-1540 steps per kilometer. Children have shorter strides.
Detailed Description
The student uses the FizziQ accelerometer to record and analyze the movements generated by walking. By first observing the absolute acceleration graph for 50 steps and then simultaneously comparing the accelerometer and pedometer data in Duo mode, the student identifies the characteristic patterns of each step and understands how the algorithm of a pedometer detects and counts steps. FizziQ's Duo mode overlays the raw accelerometer trace with the built-in pedometer count in real time, and each trial can be saved to the experiment notebook for direct comparison across walking paces.
Ready to start?
Download FizziQ and try this activity with your students.