You’re reading an older version of the Slamcore SDK documenation. The latest one is 23.04.
Requirements
Note
Due to USB connection speed issues, running the software inside a virtual machine environment is not supported.
Minimum
An internet connection is required at the start of each session to access the Slamcore software license.
To run Slamcore software, the required hardware and OS specifications are:
x64 |
NVIDIA Jetson Nano, TX2, Orin AGX, Xavier NX and AGX |
|
---|---|---|
OS |
Ubuntu 18.04 or 20.04 |
JetPack >= 4.4.1 up to L4T 32.6.1 or JetPack 5.0.2 |
CPU |
x64 (with AVX2) |
6-core @1400MHz recommended (Power mode 2) |
USB port |
>3.0 |
|
Camera |
RealSense D435i, RealSense D455, OAK-D S2 Fixed-Focus |
Listed are the minimum and recommended processors, memory and storage to run each of our SDK tools and APIs.
Tool/API |
Processor |
Memory |
Storage |
|||
---|---|---|---|---|---|---|
Minimum |
Recommended |
Minimum |
Recommended |
Minimum |
Recommended |
|
Dataset Recording |
x86 (i5) or Jetson Nano |
x86 (i5) or Jetson Nano |
2GB |
4GB |
16GB ¹ |
128GB ¹ SSD |
VIO Mode ² |
x86 (i5) or Jetson Nano |
x86 (i5) or Jetson Xavier NX |
2GB |
8GB |
16GB |
16GB |
SLAM Mode ² |
x86 (i5) or Jetson Nano |
x86 (i5) or Jetson Xavier NX |
4GB |
16GB |
16GB |
16GB |
2.5D Mapping ³ |
x86 (i5) or Jetson Nano |
x86 (i5) or Jetson Xavier NX |
4GB |
16GB |
16GB |
16GB |
ROS 2 Example ² |
x86 (i5) or Jetson Xavier NX |
x86 (i5) or Jetson Xavier NX |
8GB |
16GB |
16GB |
128GB SSD |
The table above assumes no other significant processes are consuming system resources.
[1] : Disk write speeds must be sufficient to meet requirements (see Recording Datasets)
[2] : Assuming SLAM running at 15 frames per second
[3] : Assuming both SLAM and 2.5D mapping running at 15 frames per second
ROS Wrappers
The Slamcore ROS Wrappers support
ROS1
Melodic
on Ubuntu 18.04 orROS 2
Foxy
orGalactic
on Ubuntu 20.04 (locally installed, or with Slamcore’s ROS 2 Docker image)
ROS1 Visualisation
To run both SLAM and the ROS1 visualisations (RViz) on the same system requires additional resources, we have tested this on a machine with the following hardware specification:
Core i7-7700HQ CPU @ 2.80GHz
16GB of RAM
It is possible and recommended to run SLAM and the ROS visualisations on separate, networked systems.
Recording Datasets
To record datasets they will need to be written to a drive capable of sustaining write speeds set out in the table below.
Write Speed (MB/s) |
x64 |
NVIDIA Jetson Platforms |
---|---|---|
without depth |
25 |
12.5 |
with depth |
100 |
50 |