How to Deploy Inertial Navigation Sensors in Automotive Systems

This is a sponsored article brought to you by SBG Systems.

Inertial navigation sensors (INS) benefit from advantages from both Global Navigation Systems (GNSS) and Inertial Navigation technologies. They integrate gyroscopes, accelerometers, and even magnetometers for some applications, as well as a GNSS receiver accepting Real Time Kinematic (RTK) corrections to be able to provide a centimetric position.

To reach an optimal accuracy and keep it robust, not only do we need RTK corrections from a base station or a network, but proper setup and initialization are also required.

In this article, we will go through different steps and useful tools to help the user in the setup. We will use SBG Systems' Ellipse RTK GNSS/INS embedded in a typical automotive application as an example to illustrate all mechanisms.

INS/GNSS Installation and Filter Tuning to Your Application

Inertial Navigation Systems use an advanced algorithm called the Extended Kalman Filter (EKF) to fuse GNSS data and inertial measurements. This algorithm is tunable and allows the creation of profiles taking into account application specific dynamics.

This adjustment improves the sensors' error estimation. For instance, in an automotive application, the EKF performs velocity assumptions: No lateral velocity is allowed, while wave frequency is taken into account in a marine motion profile, or high dynamics in a drone application. Offering several motion profiles to fit all applications makes it possible to offer custom tuning for all users.

Once configured, an INS/GNSS optimal installation in the vehicle ensures the best performance. Here are the critical points to follow:

• The Inertial Navigation System is rigidly fixed to the vehicle and GNSS antenna(s);
• The INS is not exposed to high vibrations. In some applications, such as motorsport, mechanical dampers may be used to mitigate vibration effects.

To go further, automotive applications' specific angle conventions are fully detailed in SBG Systems' support center, a platform specifically dedicated to explaining inertial technology.

SBG Systems Ellipse Series

The Ellipse-D is a miniature and cost-effective Inertial Navigation System (INS). It embeds a quad constellation, dual frequency, and dual Antenna RTK GNSS receiver to bring centimetric position and higher accuracy orientation in the smallest package. The Ellipse-D is part of the new Ellipse Series 3rd Generation, the most advanced miniature inertial sensors for navigation, motion, and heave sensing. See the full Ellipse Series lineup.

Level Arm Definition: A Key Measurement Now Quick and Easy

The level arm is the distance between the INS unit and the GNSS antenna(s). To properly work, the INS needs to have a rigid and fixed level arm, and measurement of the distance between the antenna(s) and the INS needs to be accurate (error

On the field, lever arm measurement could be difficult. GNSS antennas are often installed on the vehicle's roof, whereas it is recommended to have the INS as close as possible to the gravity center. Therefore, it is not possible to directly and physically measure the distance between the unit and the antennas, and measurement is taken through different steps, which can result in multiple errors.

How to simplify the lever arm measurement? Post-processing software such as Qinertia makes it possible to automatically estimate lever arms and alignment for external sensors.

The process requires making a first dry run with your final mechanical setup and a data logger with a rough estimation of level arms:

• The second step uses the PPK software to post-process recorded data.
• Qinertia will then re-estimate the level arm and suggests highly accurate values.
• Finally, you directly configure your unit with these estimated data for your mission in real-time.

Alignment Phase to Reach Optimal Performance

Once the Inertial Navigation System is displayed as Initialized in the configuration window, the system is functional but not optimal yet. A typical INS requires about 5 minutes to provide full navigation performance. This phase is named "the Alignment Phase."

The alignment phase is required to let the sensors warm up and the Kalman filter self-calibrate certain parameters, such as sensors bias. The unit performance during the complete mission depends on this critical phase. During this phase and depending on the application, some motion is recommended by the manufacturer to calibrate the INS sensors.

A good way to do so in an automotive application is to drive left and right turns, with accelerations, deceleration, full stops, and so on.

RTK Setup to Reach Centimetric Position

RTK (Real-Time Kinematic) positioning is a technique used in GNSS navigation to obtain a more accurate position up to 1cm in real-time (1cm + 1ppm). RTK uses two receivers: one base and one rover, which are located in the same area (up to 20km typically).

To enable RTK operation, your system will be composed of three major components:

1. The Rover receiver, here the Inertial Navigation System with embedded RTK ready GNSS, installed in the vehicle.
2. The Base Station (or Reference Station), a static GNSS receiver located on the field which sends corrections to the Rover via a wireless link. The base station can be installed by the user, can be part of a network of base stations (NRTK), or it can even be virtually created thanks to a network of base stations: Virtual Base Station (VBS) or Virtual Reference Station (VRS).
3. A way to communicate between the Rover and Base Station - typically a UHF, or 4G GSM modem.

Using a base station network (NRTK) brings many benefits, amid it is maintenance-free. Some networks are free to use, while some require a fee. They can be accessed online and allow you to easily set the connection with your INS, with the following steps:

• Connect to an NTRIP server and set credentials if needed;
• Configure how to send to the NTRIP server GNSS position in GGA format (NMEA);
• Configure data link with the INS/GNSS installed in the car.

Then, the network will forward corrections to the Inertial Navigation System in real-time thanks to an NTRIP Client, accessible from the INS or the embedded GNSS interface when this functionality is integrated.

For miniature inertial navigation systems, a tool called SBG data logger has been designed by SBG Systems to allow corrections as well as INS status monitoring in real-time. This is particularly useful to check if corrections are correctly received and centimeter accuracy reached.

Post-Processing for further analysis and surveying applications

Some applications do not require the navigation solution to be computed in real-time but still need centimeter accuracy. In such cases, it might be relevant to investigate how post-processing could simplify your setup.

The data link between the rover and the base station or the NRTK is not required anymore. Corrections will be integrated and used for navigation computation after the mission by a post-processing software.

The alignment phase is required to let the sensors warm up and the Kalman filter self-calibrate certain parameters, such as sensors bias. The unit performance during the complete mission depends on this critical phase.

In automotive applications, GNSS outages may happen in urban areas with significant shadowing effects due to bridges, multipath effects due to buildings, or cuts due to tunnels. If the Extended Kalman Filter compensates for a large part of these effects, a post-processing kinematic software, mostly used for high-demanding applications such as mapping, will provide an even more accurate solution.

How does it work?

An INS with a datalogger allows post-processing, which gives access to new possibilities such as processing the data backward, starting at the end of the file, and going back in time to the beginning of the file. This combination or merge of forward and backward processing with some smoothing provides a better accuracy that would never be possible in real-time acquisitions.

Additionally, such software connects to NRTK or adds RTK-based recording. Corrections are always up to date because the software consistently checks the base station stability and alerts the user if any problem has been encountered.

The Post-Processing Forward/Backward/Merge Algorithm

With post-processing kinematic software and an INS with a datalogger, it's possible to process the data backward, starting at the end of the file, and going back in time to the beginning of the file. This combination or merge of forward and backward processing with some smoothing provides a better accuracy that would never be possible in real-time acquisitions.