SLAMcore Blog

Joining the NVIDIA Jetson Ecosystem to provide Robot and Product developers with Visual SLAM ‘Out-of-the-box’

Posted by SLAMcore on Dec 3, 2020 9:39:01 AM

Today, I am pleased to announce that SLAMcore has joined NVIDIA’s Jetson Ecosystem. Through this official partnership we are offering our spatial intelligence solutions to the many robot and product designers who are using NVIDIA’s leading Jetson range of embedded compute solutions. To coincide with joining this vibrant ecosystem, we are also announcing out-of-the-box support for NVIDIA’s Jetson Xavier NX embedded System on a Module. The Jetson is the leading compute solution for autonomous robots. This is an exciting moment for SLAMcore, its customers and anyone looking to quickly add robust simultaneous localization and mapping (SLAM) capabilities to their robot and product designs.


SLAMcore’s provides industry leading location, mapping and sensing (SLAM) capabilities that are robust and able to run in real-time on commercially viable robots that operate in real-world dynamic environments. SLAMcore has optimized our algorithms to work with the most widely available and cost-effective compute/sensor hardware combinations meaning that designers can integrate SLAM quickly and cost-effectively and lets them focus on solving the other challenges they face in creating autonomous mobile robots to undertake specific tasks. Joining the Jetson Ecosystem puts SLAMcore solutions within easy reach of robot designers and developers around the world. SLAMcore is providing the first out of the box SLAM software optimized for the Jetson family. We have a fast prototype SDK for immediate use and custom capabilities for production systems and hardware.

Today it is possible to download our software to work with the Jetson Xavier NX and an Intel RealSense Depth Camera D435i out-of-the-box with no configuration needed for fast prototyping. Robot designers and developers can be up and running with one click in as little as 30 seconds. As they move to production, our algorithms can be precisely optimized to a wide range of hardware combinations to deliver robust SLAM that can cope with real-world deployments on cost-effective sensor and compute systems.

We’ve been working with NVIDIA’s excellent embedded processors for over two-years. Our initial proof of concept modules utilized the company’s hugely successful TX2 processors and we recently demonstrated our SDK working with this chipset at the Embedded Vision Conference. As a plug-and-play software company, SLAMcore’s focus is to develop leading SLAM solutions that robot designers can quickly incorporate into their designs. We have optimized our algorithms to work seamlessly and make the best use of the power of the Jetson Xavier NX. Anyone using the combination of this system on a module and the Intel RealSense Depth Camera, can be assured that our algorithms will provide robust, accurate and reliable SLAM for rapid prototyping and proof of concepts - and can be readily optimized for production systems.

NVIDIA is recognized as a leader in mobile edge AI using its world-beating graphics processors in combination with low-power CPUs to deliver the edge processing and AI capabilities demanded by robots and an array of other intelligent devices. With a 6-core NVIDIA Carmel ARM®v8.2 64-bit CPU and the 384-core NVIDIA Volta™ GPU, the Jetson Xavier NX has unrivalled performance, yet is less than the size of a credit card and consumes as little as 10 watts of power. As such it is rapidly becoming the platform of choice for robot designers the world over who can buy single units for prototypes and proof of concepts, right up to thousands of units for at-scale deployments. This mirrors our own licensing which starts with a per-unit monthly fee for development and scales to a per-robot fee for deployment.

As a member of the Jetson Ecosystem SLAMcore joins a select group of leading technology companies working closely with NVIDIA to extend the applications of its technology. Thanks to the high-quality hardware produced by NVIDIA, engineers can put the Jetson Xavier NX at the heart of their designs. The computational overhead of our SLAM algorithms is very low, and even our machine-learning that supports our perception and semantic segmentation capabilities have a small footprint on the overall power of the CPU and GPU cores in the Xavier NX, leaving plenty of space for the wide range of other processes needed to deliver a fully functioning autonomous robot.

Visual SLAM has for a long time been a complex, expensive and time-consuming element of robot design - it is however essential to effective operation in the real-world. Open-source solutions that work well in the lab are often not suited to deployment in ‘live’ rapidly changing dynamic environments. At SLAMcore we’re proud that our algorithms are among the best for visual SLAM solutions. Now, working with NVIDIA we can deliver these solutions quickly and simply to thousands of designers working to build autonomous robots. One-click download and up and running with SLAM on Xavier NX in under 30-seconds. If you’d like to see for yourself, please get in touch.



Topics: RobotsForGood, Robotics, VisualSLAM, Embedded Processors, Fast Prototyping