Using IMU to enhance robot positioning: a fundamental technology for accurate navigation

summary

This article highlights the importance of inertial measurement unit (IMU) sensors for robot positioning and outlines their key benefits. IMUs provide critical motion data and have become an essential component for precise positioning of robots. Integrating accelerometers, gyroscopes, and magnetometers, IMUs enable robots to accurately determine their orientation, position, and motion by providing real-time responses, allowing robots to navigate in dynamically changing environments. Sensor fusion technology combines IMU data with other sensors, such as cameras or LIDAR, to improve positioning accuracy by integrating multiple data sources. IMUs are widely used in mobile robots, humanoid robots, unmanned aerial vehicles (UAVs), and virtual/augmented reality. They play an important role in achieving precise positioning, allowing robots to perform complex tasks autonomously and interact effectively with their surroundings. This article explores the use cases for IMUs and how IMUs play a key role in achieving precise positioning in the challenging operating environments of AMRs.

Introduction

Autonomous mobile robots (AMRs) are essential to the smart factories and warehouses of the future, playing a key role in shaping the automated, sustainable and clean factories of the future. AMRs increase efficiency, reduce waste and optimize utilization in industrial environments. While it may be possible to build and optimize factories specifically for AMRs in the future, there are still many challenges in adapting these robots to existing warehouses and factories. The main obstacles facing AMRs involve two key parts: efficient path planning (determining the best path) and accurate positioning (continuously updating its position in the environment). 1

This article focuses on indoor navigation in closed environments where GPS coverage is not available. AMRs utilize a range of sensors and algorithms for positioning and navigation. These include vision sensors such as cameras, LIDAR, and radar, as well as odometry sensors such as wheel encoders and IMUs. Each sensor modality has its own advantages in terms of range, accuracy, and sensory information. The combination of these sensors ensures comprehensive data to effectively localize the robot in dynamic environments. While full autonomy requires a range of sensors, this article focuses on the use cases of IMUs in challenging AMR operating environments and how IMUs help achieve precise positioning, which is critical for navigation and autonomy.

What is an IMU?

IMU is a micro device composed of micro-electromechanical system (MEMS) devices. It usually includes:

►Triaxial accelerometer: An accelerometer is used to measure acceleration relative to the Earth’s gravitational field. In an IMU, a triaxial accelerometer is used to measure the x, y, and z axes (see Figure 1).

Figure 1. Acceleration measurement on the x, y, and z axes.

► Three-axis gyroscope: A gyroscope is used to measure the rate of rotation, providing the angular velocity of each of the three axes. A three-axis gyroscope can measure the angular velocity of the robot on the x, y, and z axes (ωx, ωy, ωz) (see Figure 2).

Figure 2. Gyroscope measurements on the x, y, and z axes.

►High-performance magnetometer: Provides magnetic field measurements, which are critical for accurately estimating orientation in challenging environments. Although not popular, some traditional IMUs are equipped with magnetometers.

►Others: Temperature sensors are used to compensate for temperature changes, and barometers are used to measure pressure.

IMU Functional Block Diagram

►A typical IMU includes not only gyroscopes, accelerometers, and temperature sensors, but also analog-to-digital conversion to extract the measurements and temperature compensation (see Figure 3).

►The IMU uses an onboard preliminary filtering algorithm, such as an onboard FIR (Finite Impulse Response).

►Calibration and compensation corrects any misalignment or sensor bias.

►Users can choose to rotate (dƟ) the internal axis from the IMU module to match the robot’s reference frame before transmitting the final data.

Why is an IMU beneficial for AMRs?

►Real-time positioning with high refresh rates: Autonomy and real-time navigation are key elements in robotic operating environments. However, the refresh rate of perception sensors is usually limited to a range of 10 Hz to 30 Hz. In contrast, IMUs have the ability to provide high-fidelity position outputs up to 200 Hz. Higher refresh rates significantly improve the reliability of the system when it quickly adapts to rapid changes in direction in dynamic environments, thereby facilitating rapid response. With the accelerated refresh rate, AMRs are also able to provide estimated poses in the short intervals between other measurements. Therefore, IMUs play a key role in achieving real-time positioning, with refresh rates 10 times faster than perception sensors.

►Dead Reckoning: IMUs are the backbone of dead reckoning, a navigation technique that estimates current position based on previously known positions. IMUs provide position, orientation, and velocity data over time, enabling precise estimates that help AMRs navigate reliably.

►Compact size and weight: IMUs have a compact size and lightweight design, making them ideal for integration into various mobile robot configurations. For example, the ADIS16500 from Analog Devices measures only 33.25 mm × 30.75 mm, ensuring efficient placement without compromising the robot’s maneuverability.

►Reliability in different environments: IMUs have a certain ability to resist electromagnetic interference and can operate in a variety of environments, including outdoor and indoor environments. Therefore, they are suitable for a wide range of applications.

► Improved reliability through faster refresh rates: Perception sensors are typically limited to refresh rates of approximately 10 Hz to 30 Hz, whereas IMUs offer significant advantages by providing high-fidelity position outputs with up to 4 kHz of raw data. Higher refresh rates enhance reliability, especially in dynamic environments, enabling AMRs to respond quickly and helping to estimate attitude in short intervals between other measurements.

Figure 3. Typical functional block diagram of an IMU.

Why IMU is still essential for AMR when vision sensors are already available

Figure 4. Sensor stack for an AMR.

As shown in Figure 4, AMRs typically have multiple visual sensors, such as time of flight (ToF), cameras, LIDAR, etc. Although visual odometry provides a rich data set, IMUs are still necessary. The following scenarios explore some of the reasons behind this:

1. AMR navigation in feature-sparse corridors: Simultaneous localization and mapping (SLAM) algorithms essentially work by matching observed sensor data that is stored in a map for localization within the map. When an AMR traverses a long corridor (see Figure 5), it can quickly lose its position. SLAM has difficulty accurately localizing due to the lack of unique features, such as straight walls with uniform color, texture, or reflectivity. In this case, the IMU can act as an important guidance system by providing heading and direction information.

Figure 5. AMR loses visual odometry capability in corridors with unclear features.

2. Navigation in a wide open environment: Range limitations: When working in large open spaces (such as large warehouses of 50 m × 50 m), AMRs have difficulty positioning because the individual unique features are beyond the sensor range (the maximum range of LIDAR is usually about 10 m to 15 m). As shown in Figure 6, the range measurement function of the AMR cannot be used due to the large space. In addition, warehouses usually have uniform features, which also poses difficulties for visual sensors. In this case, IMUs and wheel encoders are the only reliable sources of accurate local positioning.

Figure 6. The sensor’s limited field of view (FoV) prevents the AMR from localizing itself in wide, open spaces.

3. Driving on slopes: Traditional SLAM algorithms relying on LIDAR face challenges when driving on slopes because the 2D point data does not show slope information. Therefore, slopes are misinterpreted as walls or obstacles, resulting in higher map costs. Therefore, traditional SLAM methods using 2D systems become ineffective on slopes. IMUs can help solve this challenge by extracting slope information (Figure 7), allowing for effective navigation on slopes.

Figure 7. AMR driving on a slope.

Table 1. Pose and orientation estimation of various sensor modules used for localization

4. Environmental factors during navigation: Sensitivity to environmental factors: LIDAR sensors are sensitive to various environmental factors, such as ambient light, dust, fog, and rain. These factors degrade the quality of sensor data, which in turn affects the performance of the SLAM algorithm. Similarly, other sensor modules are also affected by reflective surfaces and dynamically moving objects (other AMRs or workers), causing further confusion for SLAM. Table 1 summarizes the impact of the environment on different sensor modules. IMUs can operate reliably in a variety of environments, making them a suitable choice for mobile robots when it comes to versatility.

However, no sensor is perfect!

Although IMUs have their advantages, they also present risks and present several challenges2:

1. Noise: IMU measurements are susceptible to noise, which can reduce the accuracy of robot navigation and control. To compensate for noise, IMUs often use advanced filtering techniques such as Kalman filtering or FIR.

2. Bias: IMU sensors accumulate bias over time, which can cause errors in orientation and motion estimates. To address this, bias estimation algorithms are used to continuously update IMU sensor readings.