A flex sensor (or bend sensor) is one of the simplest possible inputs for gesture and wearable projects: a thin strip whose electrical resistance increases as it bends, wired into a microcontroller through a basic voltage-divider circuit. No signal processing, no calibration algorithm required to get a first reading — just a resistor that changes value, which makes it a genuinely approachable entry point into gesture-sensing hardware.
Watch: Arduino Flex Sensor Glove Tutorial (Au Robots, YouTube)
Step 1: Build the voltage-divider circuit
A flex sensor alone doesn’t produce a readable voltage — it needs to be paired with a fixed resistor (commonly 10kΩ) in a voltage-divider configuration: power on one end, the fixed resistor and flex sensor in series, ground on the other end, with the microcontroller’s analog input reading the voltage at the junction between them. As the sensor bends and its resistance rises, the voltage at that junction shifts measurably — that shift is the actual signal being read.
Step 2: Read raw values before mapping anything
Connect the analog output to an Arduino analog pin and print raw readings to the serial monitor while manually flexing the sensor through its full range — flat, moderately bent, fully curled. This establishes the actual min/max range for your specific sensor and wiring (which varies somewhat sensor to sensor), rather than assuming a textbook range that may not match reality. Skipping this step and hardcoding assumed values is one of the most common reasons a first flex-sensor project reads inconsistently.
Step 3: Map raw values to something usable
Once the real min/max range is known, Arduino’s map() function converts the raw analog reading into a usable range — 0-180 for driving a servo motor’s angle, 0-127 for a MIDI control-change value, or a simple threshold for a binary “bent/not bent” trigger. This is the same basic pattern used across nearly every sensor-to-output pipeline: read raw, establish real range, map to target range, act on the result.
Step 4: One sensor per finger, for a full gesture glove
A complete gesture-reading glove typically uses one flex sensor per finger, each wired through its own voltage-divider circuit into a separate analog input — five sensors, five independent bend readings, combinable into gesture logic (all fingers curled = “fist,” specific fingers extended = a recognized sign or pose). Start with a single sensor and a single clear action before scaling to a full glove; debugging five simultaneous sensor circuits is considerably harder than confirming one works reliably first.