Inertial navigation systems (INS) play a crucial role in autonomous driving technology, especially in addressing the limitations of other sensors such as GPS, cameras, and LiDAR. It provides continuous, high-frequency, and undisturbed motion state information, and is one of the core components of autonomous driving perception and positioning.

The Core Function of Inertial Navigation System

The core function of inertial navigation system in autonomous driving is the perception of vehicle motion status. Measure the three-dimensional position, velocity, and attitude angle (including roll , pitch, yaw) of the vehicle. The inertial measurement unit (IMU), as the core sensor of INS, has a very high data update frequency (usually above 100Hz), far exceeding GPS (1-10Hz) and camera/LiDAR (10-30Hz), and can capture the instantaneous dynamic changes of the vehicle.

1.       Key application scenario: Compensating for other sensor defects

Usually in autonomous driving navigation systems, GPS signals are often lost or unreliable, causing GPS signal interruption or severe degradation, such as in tunnels, underground garages, and under elevated bridges where satellite signals are completely blocked; In urban canyons and areas with high-rise buildings, GPS signals are severely reflected and subject to multipath interference, resulting in a significant decrease or even failure in positioning accuracy; Under dense forests, leaves may also block satellite signals.

At this point, INS systems typically play an important role. Through dead reckoning, based on the known precise position and attitude at the previous moment, the acceleration and angular velocity measured by IMU are integrated to calculate the current relative displacement and attitude change of the vehicle, thereby calculating the new position and attitude. This ensures the continuity of positioning. By providing high-frequency attitude information, even when the GPS signal is good, the high-frequency, high-precision attitude information (roll, pitch, yaw) provided by INS is difficult for other sensors to provide alone.

The following table compares the navigation performance indicators of the I4500 Integrated Navigation System during satellite-assisted navigation versus satellite signal loss scenarios.

I4500 System Performance

2.       The core hub of multi-sensor fusion

The modern auto drive system adopts sensor fusion technology without exception. INS is a key node in the fusion framework.

⚪  By integrating with GNSS, a GNSS/INS integrated navigation system is formed, which is the most classic and mature combination. GNSS provides absolute position and long-term stability, but updates are slow and susceptible to interference; INS provides high-frequency, continuous relative motion information and attitude, but there is cumulative error (drift). The Kalman filter utilizes the advantages of both to mutually correct, GNSS corrects the drift of INS, and INS provides continuous positioning and smooth GNSS output when GNSS fails. Secondly, it can improve overall accuracy and robustness, and the accuracy and reliability of the combined system are much higher than those of individual GNSS or INS.

⚪  By integrating with the speedometer, the speedometer provides wheel speed information (speed, distance traveled), which can assist in correcting INS errors in speed estimation, especially when the vehicle is driving straight. The following table shows the performance indicators of the I3700 integrated navigation system produced by Micro-Magic Inc. Even in the case of satellite signal loss, high measurement accuracy can still be achieved through the algorithm fusion of INS and wheel speedometer.

I3700 Navigation Accuracy Indexs

⚪  By integrating with visual/LiDAR SLAM, the high-frequency data of IMU can provide motion prediction for visual or LiDAR processing, reducing the computational complexity of image matching or point cloud matching, and improving real-time performance and robustness (especially in fast motion or weak texture environments). The precise attitude information (roll, pitch) provided by INS is crucial for correctly analyzing the geometric relationships of camera images or LiDAR point clouds on slopes and bumpy roads.

Application Cases

Taking the I6700 product launched by Micro-Magic Inc as an example, this system can integrate various auxiliary sensors such as GNSS, Odometer, Magnetometer, etc., and provide accurate heading correction function for vehicles in various operating scenarios

I6700 Heading Correction Method

High precision inertial navigation system products launched by Micro-Magic Inc

Summary:

Inertial Navigation System (INS) is the backbone of autonomous driving positioning system. It provides continuous, high-frequency, and undisturbed vehicle motion status and attitude information, which is a key technology to ensure positioning continuity, robustness, and high-frequency response capability. Especially in challenging scenarios where GPS losing lock (tunnels, urban canyons), INS maintains its positioning capability through dead reckoning, which is an indispensable part of safety redundancy. Although its inherent cumulative error needs to be closely integrated with other sensors (GNSS, wheel speed sensors, vision, LiDAR) for correction, in the multi-sensor fusion framework, INS serves as the core hub, greatly improving the accuracy, reliability, and dynamic performance of the entire positioning system. With the advancement of IMU technology (such as the improvement of MEMS gyroscope accuracy and the miniaturization of solid-state laser gyroscopes) and the optimization of fusion algorithms, the role of INS in autonomous driving will become increasingly important.

• Предыдущий пост

  The role of a small resistor in series with the SPI bus

• Следующий пост

  Working mechanism of bootstrap capacitor of buck power chip