Problems We Solve Markets Radar Solutions Projects Why Navtech? Innovation Lab Support

Watch back the Navtech-Oxbotica Terran360 launch webinar

Published: 16/04/2021

About this Webinar

In this candid online event, hosted by Dr Inken Braunschmidt, Chief Innovation and Digital Officer at Halma, Navtech Radar’s Founder and Managing Director, Philip Avery, and Oxbotica Founder and CTO, Professor Paul Newman discuss how they joined forces to develop a new radar based localisation solution that is unlocking the full potential of automation with proven accurate localisation, even in the harshest conditions and environments.

They are also joined by Mike Woods, Chief Technologist at BAE Systems Maritime Services who talks about how they are already applying this advanced technology to bring a new level of capability to their marine operations. 

Dr Inken Braunschmidt, Chief Innovation and Digital Officer at Halma

Phillip Avery, Founder and Managing Director at Navtech Radar

Paul Newman, Founder and CTO at Oxbotica

Mike Woods, Chief Technologist at BAE Systems Maritime Services

Webinar Q&A
There were several questions asked during the webinar, here we address each of them. If you still have a question don't hesitate to contact us.
If the radar sensor cannot be attached to the roof of a vehicle, can two sensors be used instead e.g. one at the front and one at the back?

The radar can be mounted on the front, side or back of a vehicle. The system can operate as normal with a restricted field of view e.g. 130°, meaning only one radar is required per vehicle.

How do you configure the sensor on mining vehicles so that is it not sticking up above the vehicle? Will there be a 180° panel version?

You can fit the radar sensor to the front of the mining vehicle and simply just use 180° of its range. In the case of a recent train trial, a single sensor was mounted on the front of the train to provide 180° coverage.

How easy is this product to install into vehicles

It is very easy to install. You simply plug it in and it outputs position information, rather like a smart sensor. Only the most basic integration is needed, such as connecting with a power supply.

Why is this tech better and more reliable than others? Is there any case that it fails? What places it doesn’t work? What testing have you done to prove your claim it works 100% of the time?

The technology is better because it operates in all weather and light conditions, including underground, unlike GPS. The only case when it would fail would be in a place where the radar sensor gets absolutely no reflections back from its environment. Somewhere that is fully radar absorbent, for example, where it is utterly flat. One would assume that in the Arctic, for example, it would not work, but we have tested the radar in the Arctic and in other sparse environments and it has performed reliably.

What testing have you done to prove your claim it works 100% of the time?

We have tested the technology in mines, ports, on boats, cars and trains. In every circumstance with successful results.

Can it track a specific target? Are obstacle velocities available?

No, Terran360 is a localisation system, it does not do object tracking. However, clearly, radar is often used for object tracking. This will be a new product offering coming soon. If you don’t want to wait, customers have access to the standard output of our imaging radar.

Does Terran360 provide vehicle orientation (roll, pitch, yaw) information in addition to location?

It provides yaw, but not roll and pitch.

How much data is needed to be uploaded for map updates (during a highway or urban journey)?

Uncompressed, a map is less than 10 MB per km. This is low when compared to Laser, for example, which can up be up to 1 GB per km.

When used on inland waterways, do we first need to build a map of the waterways as a database to be used? Does the system need training for operation in a new environment e.g. passing a new lock the first time?

The system doesn’t need training. It does need a map, however it quickly and easily creates a multi-experience map for you, the first time it moves through your environment, without any need for external infrastructure. The system also features ego-motion estimation, so even without a map, it provides position and velocity estimation through accurate dead reckoning.

Doesn’t the system require a pre-defined map? To make the system work globally, would maps would be required globally?

Terran360 builds a map of its workplace. The maps are very small, light and quick to create. The system is also aware of how it is moving, irrespective of a co-ordinate frame, through ego-motion estimation.

How stable are the maps you create over long periods of time – do they need updating?

Mapping updates are possible but rarely needed. Radar maps are much more stable than vision and laser maps. But of course, Terran360 gives a lightweight way to update your maps should you need to.

Does this solution provide absolute or relative localisation? Is it a SLAM based solution? Does it require any prior information like a map of the environment?

It is not a SLAM based solution. However, it does produce its own maps and it can output both relative and absolute position depending on how the map is built.

Does it really not use GPS at all? How can you prove that you can work without GNSS/GPS? How is it affected by GPS jammers?

The solution operates entirely independently from GPS/GNSS. To demonstrate its capability to you, we can ground-truth the performance of Terran360 with GPS, and then run it without it. We would encourage you to contact us for a trial, so we can test its performance for your application as a proof of concept. As Terran360 operates entirely independently, its performance is unaffected by GPS jammers and is consistently reliable.

How does the SWaP, Cost and Servicing of an m-scan radar fit within autonomous vehicle markets?

It fits very well, particularly because our radar requires no servicing throughout its lifetime, and mechanical scanning not a cost driver. For Terran360°, off-road industrial automation is our primary market.

From a price-point perspective, how does the Navtech radar compare to LIDAR solutions?

As a solution, Terran360 which uses Navtech’s 360°, long-range, all-weather, high-resolution radar is cost-efficient, without you having to compromise on reliability. Set up is infrastructure-free with no need to install physical beacons and integration is simple. A single sensor solution, the radar also requires no maintenance for its product lifetime. By providing reliable, continuous localisation information in all conditions, Terran360 can enable you to unlock the potential of your autonomous operations, so you can increase safety, productivity and efficiencies to give you a competitive advantage. It is a world-first in radar localisation. As a radar, the Navtech sensor offers different capabilities to Lidar. It’s proven, it’s been out in industrial automation and running ports for 20 years.

Can this technology be used across port operations, including ship docking, warehouse operations and hinterland traffic? Is it also possible to use for military, airport, and mining applications?

Navtech and Oxbotica have carried out many proof of concepts. Using the solution on airports would be fine as there are plenty of reflections. Unless you have an area with no reflection or where everything is moving such as in the middle of the sea, then it will work. It will work across port operations, we do have customers who are looking at using our technology for the last 50 metres of ship docking which the high-resolution radar is superior to conventional navigation radar.

I’m interested in the rail application. Aside from localisation, what other information can the system provide about the environment, for example detecting changes in vegetation, landslips etc?

We have trialled our technology for rail localisation. Providing other information about the environment is not a feature of Terran360, however, it is a capability we could provide in another offering in the future. The raw output of the imaging radar is available for customers to use and do development with through Navtech’s SDK.

Have you been able to test it in a radiated or nuclear environment? Did you experience perturbances in the localisation calculations?

We haven’t tested Terran360 in this environment yet, but we’d very much like to do so. We don’t expect there to be a problem, however we always recommend that as a first step you carry out a proof of concept with the support of our technical team.

What is the localisation accuracy compared to RTK based GPS system? Have you published comparisons that could be consulted? Is a quality of service (HDOP) value available?

Absolute position is of the order of a few centimetres. Orientation is of a fraction of a degree in typical use cases. Results are available on request. Yes, confidence measures are available, but not specifically HDOP.

What are the smallest objects that we can expect to catch with this technology and at which distance? In which frequency is position data updated?

It depends on the radar cross-section of the object you are looking at. The radar sees grass and vegetation and undulations in the ground, as well as a large objects, for hundreds of metres. Each application is different and we would advise that you trial the system. The radar scans at four or eight Hz, and position updates are available at higher rates.

As the radar is on the moving device, how is the communication with the control centre to get the location of the moving device? Is there any transmitter or receiver onboard?

No, there is no transmitter or receiver on board. It is for the host vehicle to forward the data. It is not something we provide.

What governs the hardware product form factor? Is it feasible and are there plans to reduce the overall size? 

The size of the antenna is the driver of the dimensions of the hardware. Navtech will be releasing an even smaller sensor that will work with the current software in 2022.

How much computational power and hardware resources are required to run the system? Is the tracking based on machine learning?

No GPS is required, and the system currently runs on a one Watt processor.

Do you still use “volumetric probabilistic distributions” for this radar-based localisation?

The algorithms are proprietary and are key to delivering on the accuracy and reliability of this world-first localisation technology, together with the all-weather, radar technology.


Will Terran360 work on autonomous drones as well? If not, what features will need to be added to make it work for drones?

The solution is not designed for this application.


I am interested, when is the system available?

We are already working with partners on proof of concept trials for their specific applications. Please get in contact if you’d like to organise a trial.

Never-seen-before pinpoint-accurate localisation, everywhere, every time.

Performing to a centimetre accuracy, our world-leading localisation solution enables the autonomous vehicle to answer the fundamental question “Where Am I?” in even the harshest environments.