An IMU is an integrated sensor suite equipped with multiple accelerometers and gyroscopes to conduct three-dimensional measurements of specific force and angular velocity relative to an inertial reference frame. In recent years, the term IMU has become a general name covering various inertial systems including Attitude Heading Reference Systems (AHRS) and INS. An IMU itself cannot deliver navigation solutions such as position, velocity or attitude. An INS integrates IMU measurements to output navigation results.
Inertial sensors are generally classified into three performance grades:
- Marine and Navigation Grade INS: Large, High-precision and Costly Marine-grade INS represents the highest commercial sensor grade, deployed on ships, submarines and occasionally spacecraft. It can provide unaided navigation solutions with drift less than 1.8 km per day, and the cost may reach USD 1,000,000.
Navigation-grade INS delivers slightly lower performance than marine-grade products. It is commonly adopted in commercial airliners and military aircraft, with drift below 1.5 km/h and a price up to USD 100,000.
- Tactical and Industrial Grade Inertial Sensors: High Value for Mainstream Applications Tactical and industrial sensors form the most diversified category, covering a wide range of performance and cost combinations with massive market potential. This grade is widely used in mass-production scenarios requiring high-performance data at moderate cost, such as autonomous lawn mowers, delivery robots, UAVs, agricultural robots, mobile industrial robots and autonomous vessels. Despite lower cost, trade-offs must be made in accuracy, performance, size and weight.
- Automotive and Consumer Grade Sensors: Commodity Applications Automotive grade is the lowest tier for commercial IMUs. These sensors are often sold as separate accelerometers or gyroscopes. Many manufacturers combine accelerometers and gyroscopes from different suppliers to build standalone IMU units. Even integrated with GPS or other navigation systems, automotive-grade IMUs lack sufficient precision for pure inertial navigation. They are widely used in suspension systems, airbags, anti-lock braking systems, infotainment and similar applications. This category has the lowest cost, around USD 100 per sensor.
Three core factors shall be considered for sensor selection according to practical requirements:
- Precision specifications of sensors
- Sensor price and procurement accessibility
- Difficulty of hardware design
Gyroscope Selection
Key indicators for gyroscope selection are listed below:
- Measurement Range Measurement range shall be confirmed first during sensor selection. For common flight controllers and inertial navigation systems, 300°/s or 450°/s is generally sufficient. For high-precision platforms, smaller ranges are selected for better accuracy according to application scenarios.
- Bias Stability Bias denotes gyro output under zero-input conditions, converted to equivalent input angular rate from long-term mean output. It reflects the dispersion of observed values around bias, for example, 0.005°/sec means 0.005° drift per second. Long-term steady output under zero input is a stationary random process fluctuating around the mean value (bias). The mean square deviation describing such fluctuation is defined as bias stability. Initial bias error is static and time-invariant, which can be calibrated via software.
- Scale Factor Scale factor refers to the ratio between gyro output and input angular rate. It is expressed as the slope of a straight line fitted by the least square method based on measured input and output data across the full angular rate range, and can be calibrated via software.
- Angle Random Walk Under zero-input conditions, gyro output consists of white noise and slowly-varying random functions. The slowly-varying component determines bias and bias stability. White noise is defined as the standard deviation of equivalent angular velocity under the square root of unit detection bandwidth, with units of °/sec/√Hz or °/hr/√Hz. It can also be represented by the angle random walk coefficient in °/√Hz, describing accumulated output error induced by white noise. Under stable external conditions, the main statistical characteristics of the above noise remain unchanged over time.
- Vibration Resistance Most designers prioritize bias stability, which has long been treated as the core benchmark of gyro specifications, reflecting the lower limit of gyro resolution. However, practical errors often make actual performance deviate from datasheet values. Many parameters including bias, scale factor, cross-axis error and thermal drift can be compensated by calibration algorithms.
The design goal of gyro selection is to minimize dominant error sources. For most applications, vibration sensitivity creates the largest error. Other parameters can be easily improved through calibration or multi-sensor averaging, while bias stability contributes a smaller share to the overall error budget.
Ideally, gyroscopes only measure rotation rate, but all real gyroscopes have certain acceleration sensitivity. It manifests mainly as linear acceleration sensitivity (g-sensitivity) or vibration sensitivity (g²-sensitivity). Since most gyro-equipped carriers move and rotate within Earth’s 1 g gravity field, acceleration sensitivity frequently becomes the primary error source.
Low-cost gyroscopes adopt simple and compact mechanical structures optimized for cost rather than vibration resistance. Severe errors may arise under vibration. A g-sensitivity higher than 1000 °/h/g (or 0.3 °/s/g) is common, ten times worse than high-performance gyroscopes. For such devices, bias stability becomes meaningless; slight rotation within Earth’s gravity field triggers huge errors caused by g and g² sensitivity. Such gyroscopes usually do not specify vibration sensitivity, which is assumed to be very high by default.
Mechanical vibration isolators are often added in practical applications, yet their design is complicated. Their frequency response is not flat across a wide spectrum, especially at low frequencies, and damping characteristics change with temperature and service time. Therefore, vibration resistance is a critical indicator in most gyro selection processes.
Accelerometer Selection
Accelerometers measure acceleration, tilt, vibration and shock for applications ranging from wearable fitness devices to industrial platform stabilization systems. Hundreds of accelerometers with varied cost and performance are available on the market, and there is no unified industry grading standard for accelerometers.
Key parameters for accelerometer selection include measurement range, noise density, random walk, in-motion bias stability and bandwidth, which are similar to gyro parameters.
Terms such as consumer grade and industrial grade are commonly seen in datasheets. Mid-range industrial accelerometers integrated in IMUs outperform consumer-grade products by more than one order of magnitude in precision. Industrial MEMS devices undergo comprehensive characterization of all known potential error sources.
