It’s a really exciting time to be in robotics. A huge variety of companies and organisations are exploring how autonomous mobile robots (AMRs) fit into different niches and markets around the world. And one of the most exciting of those categories is what’s known as Field Robotics. Our latest blog talks about using INS technology in field robotics and how you can make the most of the technology.
With RoboBusiness coming up in just a few days, we discuss one of the key challenges in field robotics – localisation and explain how you can reap the rewards of INS technology in field robotics. If you have any follow-up questions, our partner U-blox will be at RoboBusiness this year, showcasing our latest board set INS, the xRED3000, that’s ideal for autonomy projects.
The localisation challenge in field robotics
Much of the autonomous behaviour of AMRs is only possible if the robot knows where it is. And to do that, it needs accurate localisation data.
For indoor applications, people are using a wide range of technologies, from ultra-wideband (UWB) to SLAM, LiDAR, and camera odometry. But when you go outside, some of those technologies are more challenging to implement, while others are completely unsuited to the environment.
For one, in many field robotics settings there’s nowhere to install the infrastructure needed for things like UWB localisation. And for another, many applications of field robotics require the robot to either visit locations that haven’t been surveyed before, or to operate over such a large location that surveying in enough detail is prohibitively complex and/or expensive.
In field robotics, then, nearly everyone is turning to inertial navigation systems for their localisation data. But challenges remain. Because despite its ubiquity, inertial navigation is full of quirks that can catch the uninitiated unawares.
Things to consider when dealing with INS devices in field robotics
For the avoidance of doubt, when we refer to an INS we are referring to a device that combines an inertial measurement unit (IMU) with GNSS receivers to provide both position data and data about the robot’s heading, pitch, roll, acceleration, and more. Both of these technologies are very widely used – but to get the most out of them, there are some things you need to be aware of. For instance:
1) GNSS data needs periodic corrections
Due to atmospheric interference to radio signals coming from space, GNSS position accuracy is usually in the range of 1-2 metres. To help improve on that, you need to feed correction data to your device to offset the ‘incorrect’ position. Simply put, correction data is generated by precisely surveyed base stations that can calculate the right offset for the data your robot receives from the satellites. You need to subscribe to a corrections service, such as U-blox PointPerfect, and work out how to get the corrections data to your robot (or you can set up your own base station, though that’s even more complex).
2) There are multiple GNSS constellations to get position data from
Most people are familiar with GPS, the US-owned GNSS constellation (network of satellites). There are three other major ones: BeiDou, GLONASS, and Galileo. If you’re building an AMR for a global market, it will pay to ensure your receiver can accept data from all four constellations. As well as giving you global reach, quad-constellation GNSS is more robust than relying on one or two constellations.
3) IMUs are not all created equal
For your demo robot, you may have used relatively cheap components – and if you didn’t, you’ll almost certainly be investigating them as you work out how to get the cost per unit down when your product goes into production. The problem, though, is that cheap IMUs are rarely accurate enough for field robotics. You also have the issue of calibration to consider, as you scale up. You may be able to bulk buy IMUs, but the nature of how an IMU is constructed means that each will need careful and different calibrations to deliver consistent performance.
4) You need to fuse GNSS and IMU data
An INS combines GNSS and IMU data to give what’s called a pose estimation – the system’s calculation of the robot’s position, heading, pitch, roll, and so on. That’s done using a Kalman Filter, and they can be complicated to build and refine so that they work. Without them, the system has no way of telling whether data provided by the GNSS or IMU is inaccurate and should therefore be discarded. If you use an INS for field robotics, ask about its Kalman filter.
5) An INS on its own won’t always be enough
Field robotics doesn’t always mean open-sky conditions. Consider grocery delivery robots operating in a city. The skyscrapers, tunnels, and underpasses all create GNSS blackspots and multipath errors that will cause an INS to go off course, even if it has a high-quality IMU inside it. To combat that, you will need additional sensors that are also fused into the final output, such as wheel speed sensors.
Hopefully you’ll agree that trying to build your own localisation solution from the ground up is a complex endeavour. And with that complexity, your capital vanishes faster, your time to launch gets further away, and your stress levels (probably) skyrocket.
OxTS can help
Though it may feel daunting at first, investing in a pre-built localisation solution will pay dividends later on. OxTS GNSS/INS technology, for instance, gives you a number of benefits compared to building your own solution:
We’ve built our technology with sensor fusion at its heart, so you can quickly and easily add in additional sensors to ensure a robust localisation solution.
Our Extended Kalman Filter is one of the best in the industry for identifying erroneous data, making your localisation data as reliable as possible.
We individually calibrate all our IMUs to give consistent performance at the fleet level.
That calibration, along with advanced processing algorithms, gives you high performance levels at an incredibly affordable price.
Our technology is ITAR-free, making it easy to ship abroad.
We have already created a variety of plugins and drivers for autonomy projects, including a ROS2 driver and a NVIDIA DRIVE platform plugin, so you can effortlessly plug your localisation solution into the wider robot control stack.
For autonomy projects, we have a few different models of INS that would be especially suitable. Our AV200, for instance, is a small, lightweight INS that provides centimetre-level accuracy and features a CAN interface for projects using the CAN bus.
If you’re heading to RoboBusiness, then you’ll get a chance to see our other autonomy workhorse, the xRED3000. our board set INS on the U-blox booth. Featuring a high-precision ZED-F9P GNSS module from U-blox, the xRED3000 delivers even higher accuracy than the AV200, while weighing just 20 grams and being smaller than a Gameboy cartridge. It’s perfect for integrating into field robotics platforms, giving you a way to rapidly speed up your development time and improve the functionality of your AMR.
Ultimately, partnering with OxTS could give you a market-leading product, faster than if you’d tried to build your own localisation solution.
Talk to U-blox at RoboBusiness
As we’ve said, our partner U-blox is at RoboBusiness this month. If you’re interested in using an INS in your field robotics project, or just want to get a closer look at the xRED3000, they would love to talk to you. They’ll be on booth 617 in the main hall.
In the meantime, if you’d like to learn more about what OxTS can do for autonomy projects, click here to visit our Autonomy page, or alternatively you can download our handy AMR navigation guide below.
Autonomous Robot Navigation Solution Brief
AMRs need a robust robot localisation solution; a tool that not only records the position and orientation of the robot, but also operates both indoors and outdoors.
This solution brief steps through the aspects we recommend our customers consider when deciding on their source of localisation for their autonomous mobile robots.
Read the solution brief to learn how the right robot localisation solution can help your AMR project, including the key questions you need to ask yourself before embarking on a project.
