FAQ

positioning systems

Accerion’s infrastructure-free localization technology with sub-millimeter-level accuracy gives mobile robots the ability to perform precision work in dynamic environments.

But how does it work? What are the specifications? What are the user site requirements? We answer all your questions on this page.

We’ve grouped the questions into categories: General, Size, Interface, Mapping and User Site Requirements.

General 

How does it work?

Our systems use optical sensors with built-in processing and storage to localize based on the floor. Using our advanced algorithms, the optical sensor is able to detect unique features in each section of the floor to later provide a sub-mm accurate position back to a mobile robot (AGV or AMR).

Which floors are supported?

Our sensor works best on concrete as it uses optics to pick up details in the floor to create signatures on which it can localize later when it crosses over them.

We find sometimes a shiny, epoxy floor can cause challenges; however, we recommend trying a Triton Pilot Package which features a floor qualifier to try any floor available to you.

Additionally, if you want to use our technology and would like recommendations on best floors to install or paint options, please contact us.

Which floor provides the best performance for Accerion sensors?

Accerion sensors work best in normal industrial operations with dry, swept concrete floors with or without an epoxy coating. If you give us more details about your floor, we are happy to discuss the best options for your operation.

I currently use QR markers, tape on the floor or magnetic tags to follow paths. How can I do this with Accerion sensors?

With Accerion sensors, you can map a “virtual line” on the ground, thereby creating hundreds or thousands of virtual QR markers to know your absolute localization as you drive over them. In addition, our “line-following” mode provides the vehicle pose and the closest point on this virtual line which you can use in your motion controller/navigation stack

What is the difference between the Jupiter and the Triton?

The main differences are size and localization ability.

The Triton is a much smaller, compact unit that provides absolute localization to add to an

existing navigation stack with wheel encoders for relative localization.

The Jupiter can provide both relative and absolute localization, but it comes in a larger size and at a higher cost.

How can we test your product?

For exceptional occasions, we offer a Pilot Package for customers who like to test the sensor performance first. This package includes the sensor, all cables, two days of engineering support for integration and full access to our online Knowledge Base.

You say your technology is sub-mm-accurate. How do you measure the accuracy?

We measure in repeat accuracy, so relative to the map. For instance, if you have a poor global accuracy of your map with, say, a few centimetres, you can still follow the same line each time and be within 1mm of the location you were before. We find that our customers see this repeat accuracy to be less than 1mm.

If you turn off the AMR/AGV and move it, will it work when turned back on?

After a reboot, the sensor automatically resets to position 0,0,0. So, if you would like it to turn back on and immediately work at the spot it shuts down, all you need to do is save the last known position in your navigation stack, and it can recover as if nothing happened.

Alternatively, we will have a “buffered” recovery mode which will guarantee you find your pose back as it will process the images offline until it finds a match.

What is the accuracy of the heading provided by the system?

Triton’s relative heading accuracy is 0.3% under good floor conditions.

Size

What is the size of the sensor?

     • Triton: 11cm x 11cm x 4cm

     • Jupiter: 34cm x 12cm x 12cm

How high is the unit mounted above the ground (what is the clearance)?

     • Jupiter sits 6 cm above the ground.

     • Triton sits 4 cm above the ground.

Does the sensor match signatures at slightly different heights?

     • Jupiter : +/-2 cm working range

     • Triton: +/- 0.5 cm working range

Interface

How do you interface the device?

• The standard interface uses an ethernet connection.

• The data interface is either ROS or C++ API.

Is it ROS DDS?

No, it’s ROS 1, but we plan to include ROS 2 in 2021.

Mapping

Mapping Process: can you share more about that?

Yes, so the mapping process links pictures of the floor with an external reference source.

Most customers are successful when they create their own mapping solution. We recommend creating an initial map by providing lidar as an external reference to the Triton. Then the Triton will save the position information from the lidar with images it records of the floor. The key point is it links pictures to coordinates.

Process in real time?

We map in real time at a reduced speed, so we currently map at 0.5 m/s. After mapping, Triton can operate at 1.5m/s.

What’s the map size we can make?

We measure this in distance, and the Triton can map up to 10 km.

Therefore, a grid map with spacing of 1m x 1m would allow you to map a warehouse up to 4900m2, more or less.

The map size can be upgraded, and 40 km should be available in 2021. If the area you need to map is larger, please let us know and we can discuss it in more detail.

Can you transfer maps from one device to another and still see the same repeatability?

Yes, they are transferable, and each sensor would see that repeatability accuracy.

Please note, it’s important to ensure the Tritons are mounted at the same location on all mobile robots. Otherwise, the accuracy can change.

Can you import another map from another machine?

The core principle is you need to get images and locations for those images. So, in theory yes, but typically it’s easier to map with the unit itself.

If another machine/camera is only recording images, this cannot create a map. Only a Triton can make a map. However, these maps can be shared with other Tritons.

User Site Requirements

How do you deal with debris or deterioration?

We compare the pictures to a fingerprint. Say you have a cut on your finger, and you go to the airport security scanner. The scanner still recognizes your fingerprint and you pass through.

We work the same way, so there is little concern with small changes.

We can also detect when it degrades too much. If there is too much degradation, you can just remap and update that area.

How does it detect when it degrades too much?

Based on the developer logs files of the sensor, Accerion can provide a “heat map” that shows where your sensor:

• Drove

• Corrected drifts

• Size of the drift correction

• Distance since last drift correction

Accerion will develop a new way of logging data for Triton, which will allow users to access this information too.

Is it possible to automatically detect a changed ground condition and to update the ground characteristics database in this case?

We are working on this. Please contact us for more information.

Can the ground features database be distributed to an AMR/AGV swarm via a server?

Yes, this can be fully automated via your fleet manager with our API. Additionally, you can transfer maps manually via our GUI between sensors.

Would dust on the floor affect the signature being read?

It depends on the size, amount and floor quality. Also, line following is better to manage this than a grid since there are more points to check. Missing a few isn’t an issue.

Get in touch!

If you have questions we didn’t answer here, please get in touch. We are happy to discuss ways to implement Accerion technology in your high performance operations. sales@accerion.tech.