A MEMS accelerometer is a tiny motion sensor built into a silicon chip, and it is widely used in robotics to sense movement, tilt, vibration, and impacts. MEMS stands for microelectromechanical systems, which means mechanical parts are made at microscopic scale using chip manufacturing methods. In a robot, this sensor helps estimate orientation, detect falls, stabilize motion, and support navigation.
Its main idea is simple: acceleration causes a small internal mass to shift, and the chip converts that shift into an electrical signal.
Inside the sensor, a suspended proof mass is held by very small springs and surrounded by fixed capacitive plates. When the robot accelerates, the proof mass lags behind because of inertia, stretching or compressing the springs by a tiny distance. This deflection changes the capacitance between the moving mass and the fixed plates, and electronics convert that change into acceleration.
Because gravity also acts like an acceleration of about 9.8 m/s^2, the sensor can detect which way is down when the robot is still or moving slowly.
Understanding Robotics: MEMS Accelerometer
The sensing structure is made using layers of silicon, metal, and insulating material. Etching removes selected material to free the moving parts while leaving anchor points fixed to the chip. Many designs use comb-shaped fingers rather than simple flat plates.
As the proof mass moves sideways, the overlap or gap between these fingers changes. A matched pair of capacitors is often used. Motion makes one capacitance increase while the other decreases.
Comparing the pair rejects some effects from temperature changes and electrical interference. This differential measurement helps the chip detect extremely small motion.
The raw reading is not a perfect record of the robot's motion. Every accelerometer has bias, which is an offset present even when the sensor is held still. Its scale can be slightly wrong, so a reading may be a little larger or smaller than the real acceleration.
Noise produces rapid random changes in the output. Temperature can shift bias and scale during operation. Calibration measures these errors and corrects them in software.
A simple classroom calibration places the sensor still in several known orientations, with each axis pointing upward or downward in turn. The results give useful estimates for offsets and sensitivity.
Tilt sensing works best when the robot is stationary or moving gently. In that case, gravity provides a steady reference direction across the three sensor axes. During fast travel, the sensor cannot separate gravity from acceleration caused by motors, turns, bumps, or braking.
A wheeled robot leaning forward while accelerating may produce readings similar to a robot tilted forward at rest. For this reason, practical robots often combine accelerometer data with a gyroscope.
The gyroscope tracks turning rate over short periods, while the accelerometer provides a longer-term gravity reference. Filtering combines their strengths and reduces their weaknesses.
Sampling rate matters because motion changes over time. A balancing robot needs frequent readings so its controller can react before it falls. A vibration monitor may need an even higher rate to capture fast shaking from gears or motors.
Sampling too slowly can make a rapid vibration appear as a false lower-frequency motion, an effect called aliasing. Engineers use filters to remove unwanted high-frequency signals before sampling, then smooth noisy data afterward. Students should check the sensor range, stated in units of g, before testing.
A low range gives finer detail for gentle tilt, but strong impacts can exceed its limit and produce clipped readings. Mounting matters too. A loose sensor board can vibrate differently from the robot frame, creating data that describes the mount rather than the robot.
Key Facts
- Newton's second law connects force and acceleration: F = ma.
- A spring-like MEMS suspension follows Hooke's law for small deflections: F = kx.
- Combining force and spring motion gives acceleration from deflection: a = kx/m.
- Capacitance for parallel plates is approximately C = epsilon A/d, so changing plate spacing changes the signal.
- At rest on Earth, an accelerometer measures proper acceleration due to support against gravity, about 1 g = 9.8 m/s^2.
- A 3-axis accelerometer measures acceleration components ax, ay, and az, with total magnitude a = sqrt(ax^2 + ay^2 + az^2).
Vocabulary
- MEMS
- MEMS are microelectromechanical systems that combine tiny mechanical structures with electronic circuits on a chip.
- Proof mass
- A proof mass is the small suspended mass inside an accelerometer that moves slightly when acceleration occurs.
- Capacitive sensing
- Capacitive sensing measures changes in electric capacitance caused by changes in distance or overlap between conductive plates.
- Inertia
- Inertia is the tendency of an object with mass to resist changes in its motion.
- g-force
- A g-force is acceleration expressed as a multiple of standard gravity, where 1 g is about 9.8 m/s^2.
Common Mistakes to Avoid
- Treating the proof mass as if it moves a large visible distance is wrong because MEMS deflections are usually microscopic and detected electronically.
- Confusing velocity with acceleration is wrong because an accelerometer measures changes in motion, not speed itself.
- Ignoring gravity in accelerometer readings is wrong because gravity appears as a 1 g signal when the sensor is supported at rest.
- Assuming one axis tells the full motion is wrong because robotic motion usually requires combining x, y, and z axis readings.
Practice Questions
- 1 A MEMS proof mass is 2.0 x 10^-8 kg and its spring constant is 0.40 N/m. If the mass deflects 4.9 x 10^-7 m, what acceleration does the sensor measure using a = kx/m?
- 2 A robot accelerometer reads ax = 0 m/s^2, ay = 0 m/s^2, and az = 9.8 m/s^2 while standing still. What is the magnitude of the acceleration vector, and how many g is this?
- 3 A stationary robot is tilted so that the accelerometer readings shift from mostly the z-axis to partly the x-axis. Explain how the MEMS accelerometer can use this change to estimate the direction of gravity.