Jotrin Electronics
Описание Количество Общий (USD) Операция
Корзина продуктов
Корзина продуктов : 0
Дома > MEMS / sensing technology > INS and Autonomous Mobile Robots

INS and Autonomous Mobile Robots

Время обновления: 2022-08-20 15:52:08


Navigation is a hot topic that never goes out of fashion for robots with mobility functions. Compared to cars with self-driving functions, self-driving autonomous navigation for robots is limited by the size and cost of the body, and the sensors applied to it need to control both size and cost.

The main genres of mobile robots in path planning navigation are vision V-SLAM and LiDAR LiDAR-SLAM, both of which build virtual maps to provide navigation for robots. However, SLAM technology alone is insufficient to meet navigation and positioning requirements. As a very simple example, without IMU and using only SLAM, once the robot does not start from its map-building origin, the robot will lose its positioning and needs to move in a random direction to track a large number of map markers to reposition itself in a specific location in the map.

IMU, so the robot can find its position

A robot capable of autonomous navigation must know its position in real-time, even if it does not start every mission at the map coordinate origin. When a mobile robot measures its position, it cannot do without the absolute angle value, which the IMU provides.

It can be said that the accuracy of the robot's overall odometer and its ability to flexibly complete position positioning in any situation has a lot to do with the IMU. Even if SLAM technology is not used, it is still necessary to use the heading projection method for positioning and navigation, i.e., by measuring the wheel rotation angle and combining the inertial measurement value of the IMU with the object detection of the ToF sensor to complete the positioning.

The importance of IMU in robot navigation systems is evident. The fusion of IMU becomes an effective means for V-SLAM solutions and laser SLAM solutions to complement their positioning shortcomings. Accurate positioning and motion accuracy is the key to efficient autonomous movement, and the feedback detection mechanism provided by IMU is very useful for optimizing the performance of navigation systems.

IMU has also evolved to integrate MEMS devices such as accelerometers, gyroscopes, and magnetic sensors to assist machine motion in a smaller size and at a lower cost. These six-axis motion sensors acquire the robot's roll, pitch, and yaw motions from linear and rotational angles. Then, combined with motion and room maps, they can pinpoint the robot's position in real space, even if it does not start from the map's origin, and quickly know where it is in real space.

An IMU-based INS localization process is not complicated. Starting from collecting multiple sets of accelerometer and gyroscope data, the INS first determines whether the attitude angle changes by calculating the quaternion and rotation matrix (if the quaternion and rotation matrix change, it involves using Kalman filtering to calculate the optimal attitude angle), then calculates the 3D acceleration in the coordinate navigation system, the 3D linear velocity, and then The introduction of INS provides a very efficient way to improve the overall navigation system orientation estimation and overall accuracy.

How robotic INS is developed

As mentioned above, the IMU-based INS has evolved to include more sensing than just acceleration sensors and gyroscopes. One of the reasons is that with the development of robotics, there is no way for a single sensor to meet the increasingly accurate navigation and positioning needs, and the trend is to integrate sensing. Second, although IMU has a unique advantage in capturing motion in multiple degrees of freedom, the unique self-reliance of IMU means that it does not solve drift, noise, and zero-offset instability well on its own.

Many sensing can be fused with INS, and the integrated magnetic sensing mentioned above is a viable option, mainly to complement IMU for target attitude information acquisition. Magnetic sensing provides the coordinate system for navigation. The robot can detect the relative change of the magnetic field around the target to determine the passage and passing of the target and complete target detection. This fusion scheme is small and inexpensive, and the specific navigation accuracy depends on the key parameters of the fused device and the algorithm. The fusion with GNSS enhances the positioning of high-precision GNSS after the loss of the observed target so that the position of the dynamic target is no longer lost. This fusion is mainly for position detection applications with many outdoor obstructions.

Whether fused or not, the sensor's parameters are the basis of the navigation system. Sensitivity error, sensitivity temperature coefficient, zero-rate offset, zero-rate offset temperature coefficient, and noise density are the most basic and important considerations. Sensitivity error and noise density are decisive parameters in robotic applications and any sensor application. In applications where the ambient temperature varies greatly, the sensitivity temperature coefficient and the zero-rate offset temperature coefficient also determine the device's suitability.

The device has to have as much support as possible provided by the algorithm, even if the parameters are excellent. The bias of the IMU is difficult to cancel even if it can be measured accurately, and this, combined with noise, can cause the navigation distance to drift. Without algorithmic optimization, the zero drift of the INS bottom sensor for 1 hour is usually more than 10°, and then the cumulative error of random motion can quickly dissipate. Combined with the kinematic model to give the system more algorithmic rules can significantly optimize the cumulative error of INS, which is also one of the core competencies of companies doing INS, doing the overall navigation of mobile robots. The sensor hardware level can open up a big gap. The algorithm level can also open up competitiveness.

Manufacturer robot IMU application

We pull back the attention to the hardware, IMU manufacturers in robotics applications of hardware strength.

Semtech Corporation

Semtech's SH3011, SH3001, and SH2100 biased consumer applications. The IMU445 multi-axis fusion inertial measurement module for typical robotics applications is industrial and automotive grade push. IMU 445 is a complete inertial system, including a three-axis gyroscope and a three-axis accelerometer, but also combined with Senodia's self-developed algorithms to provide excellent dynamic performance to complete multi-axis inertial sensing.


IMU445 gyroscope dynamic range ±250dp, initial sensitivity 0.015dps/LSB, initial offset error ±0.5dps, zero offset instability 0.0028dps; accelerometer dynamic range ±4g, initial sensitivity 0.12mg/LSB, initial offset error ±10mg, zero offset instability 0.5 mg.

Shinner Sensing

The portfolio of inertial systems is rich, from triple redundant 6 DOF IMU to embedded multi-axis inertial measurement systems to INS with GNSS integration. Its products are mainly for automotive applications, so they are more than enough for robots.


The product selected here is OpenIMU330B, a 6 DOF IMU with triple redundancy, gyro range ±400 deg/sec, zero bias stability 1.5 deg/hr, bandwidth 5-50Hz, temperature error not more than 0.3 deg/sec, acceleration range 8 g, bandwidth 5-50Hz.


As a well-known IMU manufacturer in robotics applications, TDK has a wide range of IMU product lines, from 6-axis IMUs to 7-axis IMUs that combine air pressure sensors and 9-axis IMUs that combine 3-axis compasses. TDK's main product in robotics applications is the ICM-42688-P, a high-performance 6-axis IMU for robot motion tracking applications.

The gyroscope and accelerometer noises are 2.8mdps/√Hz and 70µg/√Hz, respectively, and the sensitivity error is ±0.5% for both. The gyroscope zero-rate offset ±0.5dps and the accelerometer zero-acceleration offset ±40mg, both at the on-board level.


Bosch is the industry leader in consumer IMUs, and its BMI series IMUs have been iterated from BMI055 to BMI270 in several versions for many different application scenarios. There is also a dedicated series for robotics applications. The BMI088 high-performance inertial measurement unit is a 6-axis sensor developed for UAV and robotics applications.


The BMI088 integrates a 16-bit gyroscope and 16-bit accelerometer in a small 3×4.5×0.95mm3 LGA package. The gyroscope technology and low TCO accelerometer design used in the series are automotive grades tested, with bias stability below 2°/h and TCO below 15mdps/K. The accelerometer has a wide measurement range of ±24g with spectral noise not exceeding 230µg/√Hz. The BMI088 is capable of accurate steering and positioning in high vibration applications.


ST's NEMO inertial module integrates complementary sensors in a compact, rugged, easy-to-assemble IMU. Among the 20 part numbers, the ISM330DLC and ISM330DHCX are suitable for robotic applications, not counting automotive applications, and both maintain excellent stability as system-level packages in terms of process and packaging.

The ISM330DLC gyroscope noise density is 3.8mdps/√Hz, and the accelerometer noise is 75µg/√Hz, while the ISM330DHCX gyroscope noise density is 5mdps/√Hz, and the accelerometer noise is 60µg/√Hz. Power consumption.

Final words

INS is only an integral part of this, and the IMU's sensitivity error and noise density largely affect the overall navigation system positioning accuracy. A high-quality IMU fused with other sensors can significantly reduce the uncertainty gap in positioning.

Предыдущий: Everything about N-channel hexfet power mosfet-IRF3205

Следующий: Current detection scheme for ACS712 current sensor

Ratings and Reviews



мой профиль


Онлайн консультация