Sudo dpkg -i cuda-repo-l4t-r21.1-6-5-prod_6.5-14_bĭownload and install the CUDA Toolkit itself, including the OpenGL toolkit for NVIDIA:Īdd the user ubuntu to the group “video” to provide access to the GPU:Īdd the paths to the CUDA installation folder to the. Install the repository metadata for CUDA for L4T, which has just been downloaded: Download the deb file for CUDA Toolkit 6.5 for L4T from here (instruction taken from here ): To work with the ZED camera, we need the ZED SDK, which requires CUDA version 6.5 installed. To solve the problem, open the / etc / network / interface file from under the root: There may be a problem with connecting to Jetson over SSH: ifconfig displays the IP address, but it is not possible to connect to Jetson at this address. For authorization, use the login and password ubuntu.įind out the IP address of the Jetson host: So the Ubuntu 14.04 system will start automatically. I got NVIDIA Jetson on my hands with JetPack already installed and Ubuntu 14.04, so here I am not considering its installation. To install the system on NVIDIA Jetson, you may need to install JetPack TK1 (instructions can be found here ). Ĭonnect NVIDIA Jetson TK1 using HDMI to the monitor, connect the Ethernet cable to the appropriate connector, and finally use the 12 V AC / DC adapter to the power supply. More information about the camera can be found on the official page. The camera can be used to create outdoor maps on the drone. The advantage of the camera can be considered its small size (175 x 30 x 33 mm) and lightness (159 g). The camera is relatively inexpensive (costs $ 449). Unlike active cameras (such as ASUS Xtion and Microsoft Kinect), the ZED stereo camera does not have an IR laser to measure distance. The ZED stereo camera is a passive depth camera consisting of two ordinary RGB cameras 12 inches apart from each other, with an overview of up to 20 meters. By default, NVIDIA Jetson TK1 is preinstalled with Ubuntu 14.04 operating system.Ī 12 V AC / DC adapter is used to power the microcomputer. NVIDIA Jetson TK1 internal memory is represented by a 16 GB eMMC module. More details about all the features of the platform can be found here. As stated on the official pageThe processor allows you to run on the board any demanding graphics tasks, such as face recognition, augmented reality, and even computer vision tasks for unmanned vehicles. The mobile Tegra K1 processor has very similar properties and architecture to the desktop GPU, which allows it to withstand heavy loads with minimal power consumption. NVIDIA Jetson boasts a 4-core ARM Cortex-A15 CPU with 2.3 GHz, has the same components as the Raspberry Pi (HDMI, USB 2.0 and 3.0, Ethernet), as well as PC-specific modules: SATA, mini-PCIe. NVIDIA Jetson TK1 is an NVIDIA microcomputer based on Tegra K1 SOC (CPU + GPU on a single chip with CUDA support). To begin, consider the NVIDIA Jetson TK1 platform. After installing all the necessary, we will connect and try to strealabs ZED stereo camera. In this article I will deviate a little from the topic and talk about setting up and working with ROS on an NVIDIA Jetson TK1 microcomputer. Good afternoon, dear readers! In the last articles, I talked about popular SLAM methods and visual odometry that have support in ROS.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |