Skip to content

README.md

Latest commit

 

History

History
273 lines (210 loc) · 11.4 KB

README.md

File metadata and controls

273 lines (210 loc) · 11.4 KB

Static Badge

This page is in preparation for the release 1.0.0. See the ADCAM tag v0.1.0-a.1 for the latest release.

License

Platform: NVIDIA Jetson Orin Nano OS: NVIDIA JetPack 6.2.1

Platform: Raspberry Pi 5 OS: Raspberry Pi OS 64-bit

C++ Python

License

ADCAM Camera Kit

Latest Release

Overview

This repository contains the source code for the ADI ADCAM Camera Kit, which is built around the ADTF3175D Time-of-Flight (ToF) Mega-Pixel imager and the ADSD3500 Depth ISP.

The ADCAM hardware interfaces with the NVIDIA Jetson Orin Nano Developer Kit or the Raspberry Pi 5 over MIPI CSI-2 for image data, and uses USB-C solely for power. Unlike the earlier ADTF3175D Evaluation Kit, the ADCAM introduces two key improvements:

  • Dual ADSD3500 Depth ISPs
    Depth computation is fully handled by hardware, eliminating the need for proprietary SoC-based depth processing libraries. The only exception is the radial-to-XYZ (point cloud generation) step, which is implemented in the open-source libaditof library.

This repository depends on the following components:

Repo Purpose
ToF-drivers Provides the V4L2 camera sensor driver for the ADSD3500 Depth ISP, along with device tree sources and kernel patches as required.
libaditof Provides the SDK supporting the ADCAM system, integrating ADSD3500 Depth ISP processing with the ADI ToF imager.

Supported Platforms

Platform
NVIDIA Jetson Orin Nano Developer Kit ADCAM with Jetson Orin Nano Dev Kit
Raspberry Pi 5 ADCAM with the Raspberry Pi 5

Requirements

  • NVIDIA Jetson Orin Nano Developer Kit: JetPack 6.2.1
  • Raspberry Pi 5: Raspberry Pi OS Full (64-bit) Debian Trixie, release 2025-12-04

Examples

Core Examples

Example Language Description
tof-viewer C++ Graphical User interface for visualising stream from depth camera
data-collect C++ A command line application that takes in command line input arguments (like number of frames, mode to be set, folder location to save frame data) and captures the frames and stores in path provided
first-frame C++ A C++ that example that shows the steps required to get to the point where camera frames can be captured.
first-frame
Python 		A Python that example that shows the steps required to get to the point where camera frames can be captured.
streaming example
Python 		A Python example that shows streaming depth frames.

Other Examples

Example Language Description
ROS2 Application C++ A more extensive ROS2 example based on the ADI ToF SDK.
Stitching Algorithm C++ A stiching algorithm using ADI ToF data.

Directory Structure

Directory Description
apps Applications specific to various targets and hosts. Currently contains the server applications for streaming to the host.
ci Useful scripts for continuous integration
cmake Helper files for cmake
dependencies Contains third-party and owned libraries
doc Documentation
examples Example code for the supported programming languages
libaditof (submodule) Submodule with SDK source code
scripts Useful development scripts
sdcard-images-utils Linux image build tools
tools Standalone applications
ToF-drivers (submodule) ADSD3500 V4L2 Camera Sensor device driver

Building the Eval Kit

Note, prior to committing to the repo it is important to format the source code, see the code formatting document.

Steps below:

  1. Installing the Pre-requisites
  2. Cloning the Repo
  3. Building and Installing the Kernel Pieces
  4. Building the Eval Kit

Requirements:

  • An internet connect is mandatory.

1. Installing the Pre-requisites

Pre-requisites

  • CMake
  • g++
  • Python 3
  • OpenCV - for the examples
  • OpenGL - for the examples
  • Doxygen - for documentation generation
  • Graphviz - for documentation generation

Installing the pre-requisites

NVIDIA Jeston Orin Nano Dev Kit with JetPack 6.2.1
sudo apt update
sudo apt install cmake g++ \
     libopencv-dev \
     libgl1-mesa-dev libglfw3-dev \
     doxygen graphviz \
     libxinerama-dev \
     libxcursor-dev \
     libxi-dev \
     libxrandr-dev \
     python3.10-dev
Raspberry Pi OS Full (64-bit) Debian Trixie, release 2025-12-04
sudo apt update
sudo apt install cmake g++ \
     libopencv-dev \
     libgl1-mesa-dev libglfw3-dev \
     doxygen graphviz \
     libxinerama-dev \
     libxcursor-dev \
     libxi-dev \
     libxrandr-dev \
     python3.13-dev

Obtaining and Installing the Depth Compute Library

In addition the depth compute libraries are required.

You can get the two libraries from the ADCAM release software, but please note in which case it is under an evaluation license.

For a non-eval license please contact us at tof@analog.com.

The two libraries files are:

  • libtofi_compute.so
  • libtofi_config.so

There are two options for pointing these libraries:

  1. Set the LIBTOFI_LIBDIR_PATH enviroment variable to the path with these files.
  2. Place the libraries must be in a folder called libs that in one level below the cloned ADCAM repo folder. For example:
(aditofpython_env) analog@analog-desktop:~/dev/ADCAM$ pwd
/home/analog/dev/ADCAM
(aditofpython_env) analog@analog-desktop:~/dev/ADCAM$ tree ../libs
../libs
├── libtofi_compute.so
└── libtofi_config.so

2. Cloning the Repo

git clone https://github.com/analogdevicesinc/ADCAM.git
cd ADCAM
git submodule update --init
git checkout main
pushd libaditof
git checkout main
popd
pushd ToF-drivers
git checkout main
popd

3. Building and Installing the Kernel Pieces

Updating Linux with the ToF pieces is required before the eval kit is built. To build the kernel you will need a connection to the Internet.

If you are building on an SSD make sure the SSD has airflow since it is beneath the Jetson Orin Nano Dev Kit and may over heat as a result.

The example below using file name NVIDIA_ToF_ADSD3500_REL_PATCH_08Apr26.zip and path of NVIDIA_ToF_ADSD3500_REL_PATCH_08Apr26. Substitute with the NVIDIA_ToF_ADSD3500_REL_PATCH_* created by your build.

Note, this will take sometime to build.

NVIDIA Jetson Orin Nano Dev Kit with JetPack 6.2.1

Building & Installing
cd ADCAM/sdcard-images-utils/nvidia
./setup.sh
./runme.sh 7.1.0 main
unzip NVIDIA_ToF_ADSD3500_REL_PATCH_08Apr26.zip
cd NVIDIA_ToF_ADSD3500_REL_PATCH_08Apr26
sudo ./apply_patch.sh
sudo reboot

Raspberry Pi OS Full (64-bit) Debian Trixie, release 2025-12-04

Building & Installing

Note, this will take sometime to build.

cd ADCAM/sdcard-images-utils/rpi
./setup.sh
./runme.sh 7.1.0 main
cd NVIDIA_ToF_ADSD3500_REL_PATCH_08Apr26
sudo ./apply_patch.sh
sudo reboot

4. Building the Eval Kit

Build

On Device Build: Jetson Orin Nano Dev Kit or Raspberry Pi 5

Let's start with a standard build. Where we need:

  1. Clone the repo.
  2. Update the required submodules.
  3. Check out the requried branch.
  4. Build the code.
cd ADCAM/
mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
cmake --build . -j 6
CMake options

There are a number of build options available via the root CMakeLists.txt file: https://github.com/analogdevicesinc/ADCAM/blob/6e5b722b5c36923065c4a3be96ad0553d387e699/CMakeLists.txt#L20C1-L24C109

option(WITH_EXAMPLES "Build examples?" ON)
option(WITH_DOC "Build documentation?" OFF)
option(WITH_PYTHON "Build python bindings?" ON)
option(WITH_NETWORK "Build network interface?" OFF)
set(WITH_PLATFORM "AUTO" CACHE STRING "Platform selection") # Options are: "AUTO", "NVIDIA", "RPI" or "HOST"
  • WITH_EXAMPLES: Builds all examples that are in the examples folder. Default: ON - ie, build examples.
  • WITH_DOC: Builds the doxygen documentation. Default: OFF - ie, do not build documentation.
  • WITH_PYTHON: Builds the Python bindings library, which is required for the Python examples (see: examples/bindings/python). Default: OB - ie, build Python bindings.
  • WITH_NETWORK: Its complicated, ignore for now. Default: OFF - ie, do not build with network functionality enabled.
  • WITH_PLATFORM: Sets the target platform for the build. Default: AUTO - ie, attempt to auto detect the platform.
    • AUTO: Auto detect between the NVIDIA, Raspberry or Host device
    • NVIDIA: Forces a build for NVIDIA Jetson Orin Nano Dev Kit
    • RPI: Forces a build for the Raspberry Pi 5
    • WSL2: Forces a build for Windows WSL2 (this is only for a networked setup)

An example build showing how to change an option during the build process. For this we will disable building the Python bindings.

Starting in the root of the cloned ADCAM folder:

cd ADCAM
mkdir build
cd build
cmake -DWITH_PYTHON=OFF -DCMAKE_BUILD_TYPE=Release ..
cmake --build . -j 6