Next-generation inertial measurement technology can provide more accurate positioning and determine the position of autonomous vehicles with centimeter-level accuracy

With the continuous development of science and technology, autonomous driving technology is becoming more and more mature. Vehicles navigate autonomously by sensing the world around them, but all this is based on the commitment and guarantee of vehicle positioning accuracy and safety. In order to realize the various complex functions of autonomous vehicles, accurate and reliable location information has become indispensable.

The crowded urban canyon environment poses a huge challenge to the sensor array of autonomous vehicles. When the autonomous vehicle attempts to turn left after losing GPS/GNSS signal, IMU technology begins to play an important role.

As the main source of information for active decision-making in autonomous driving, perception sensors can perceive the world around the vehicle, including radar, lidar, infrared, ultrasonic and camera vision, but all of these require powerful computing power as a support.

Navigation systems tell autonomous vehicles where they are and where they need to go. These systems consist of GNSS/GPS receivers and INS (Inertial Navigation System), which includes inertial motion sensors and inputs from odometry and steering sensors.

Eight common sensors used in self-driving cars

MEMS-based inertial sensors such as gyroscopes and accelerometers have long been used as discrete components for collision detection, airbag deployment, and electronic stability control in vehicles. High-end IMUs (inertial measurement units) using MEMS or fiber optic technology are commonly used in aircraft and tactical guidance systems, and their performance is 10 to 1,000 times higher than traditional MEMS sensors.

As autonomous vehicle technology and safety standards advance and improve, the positioning accuracy required for IMUs and INS is approaching the standards for aerospace and tactical-grade equipment - unified and reliable centimeter-level accuracy rather than meter-level accuracy.

Until now, IMUs that can achieve this level of performance and safety have been too expensive for high-volume markets such as automotive. But now we are seeing innovations in design and manufacturing that are making high-performance IMU technology more affordable for a wide range of automation and industrial applications.

Why IMU?

An IMU is an electronic module that integrates multiple inertial sensors to generate acceleration and angular velocity measurements along multiple axes or degrees of freedom. A six-degree-of-freedom (DOF) IMU consists of a three-axis gyroscope and a three-axis accelerometer. Combining the measurements from these sensors over time using an extended Kalman filter (EKF) allows for highly accurate position, velocity, attitude, or orientation calculations. An attitude and heading reference system (AHRS) combines magnetometer readings with IMU data to calculate heading, roll, and pitch. An INS adds GPS to track an object's position, orientation, and velocity.

In a typical autonomous vehicle application, the INS combines traffic routes, high-definition maps, and a perception sensor system to determine the vehicle's route and how to navigate. When all systems are operating normally in good environmental conditions and satellite coverage, an INS with a traditional automotive-grade IMU typically provides sufficient positioning accuracy and reliability to ensure safe operation.

However, external conditions are often less than ideal. Often, GPS signals are lost or weakened due to urban canyons, tunnels, overpasses, multipath errors, or poor satellite coverage. In addition to this, vehicles may encounter other challenging conditions, such as precipitation or reflective surfaces, which may affect the performance and integrity of the camera, lidar, or radar system data.

When other sensor data used for navigation is abnormal or interrupted due to environmental reasons, the navigation task will turn to rely on the dead reckoning system. At this time, the autonomous driving system mainly relies on IMU, wheel speed sensors and visual sensors. At this time, it is particularly important to use the right level of IMU, which can continuously sense and provide position information regardless of environmental conditions.

The inherent bias and drift errors in MEMS devices will bring a certain burden to the system and need to be eliminated in time. In addition, the errors that are not compensated will accumulate into position errors, and high-end IMUs that have been strictly calibrated over temperature and time can greatly reduce these error sources. IMUs with built-in redundancy can provide more accurate position estimates and bring higher safety, integrity and reliability to the entire autonomous vehicle system and sensor fusion network.

Real-time dynamics

Another notable development in precision INS positioning is the emergence and development of GNSS RTK (Real Time Kinematic) technology. RTK, when properly fused with IMU data, can improve GPS positioning accuracy by 100 times, from meter-level accuracy to centimeter-level accuracy. RTK technology refines the position data received from GPS signals by eliminating ionospheric and tropospheric delays, multipath, satellite clock and ephemeris errors (errors caused by the GPS receiver using the position of the satellites in the position calculation). RTK systems use a survey-grade base station that broadcasts corrections to a rover (a moving object or vehicle) via a cellular signal. The correction data is fused with GPS and IMU data through complex algorithms and Kalman filters to output the precise position of the rover in real time.

RTK positioning and localization technologies are valuable for a wide range of autonomous driving applications including agriculture, construction, robotic delivery, drones and consumer autonomous vehicles.

Until now, RTK and other similar services have been expensive and time consuming to acquire. As a result, they have been used primarily in agriculture, land surveys, and construction applications. The increasing popularity of autonomous vehicles and their need for precise positioning has given rise to a new breed of RTK software that is scalable across geographies, affordable, easy to integrate, and optimally complements the sensor fusion positioning and navigation capabilities of autonomous vehicles.

For any self-driving car, it is crucial to accurately understand its location and surrounding environment, its destination, and how to get there. Although it is still uncertain when driverless cars will become consumer products, this technology has been widely used in some industries. Combined with advanced IMU and RTK, it is expected to further improve and perfect the system performance, safety and integrity, and make inertial navigation systems popular sooner.