Robot IPC: A Lightweight Inter-Process Communication Library

Motivation

Traditional solutions for multi-process robotics, such as ROS, are often overly complex for simple projects and introduce unnecessary overhead. This library was created to address the shortcomings of ROS and similar frameworks.

• Complexity: ROS is a rigid framework that forces developers to structure code within its package system, which can lead to conflicts with other libraries.
• Performance: ROS sacrifices speed and low latency to support distributed communication across a local area network (LAN), features that are often not needed for a single robot's on-board computing.
• Usability: ROS can be confusing and problematic with alternating network connections (wired and wireless). For example, a system booted with a wired connection may lose topic visibility after the cable is unplugged, even if data should be handled locally.

Similarly, many DDS frameworks, like FastDDS, are designed for distributing data across a reliable industrial network. However, for many robotic applications, a publish-subscribe model isn't required. We simply need to read the latest data when necessary, rather than tracking every single update. This is like checking a weather app for the current forecast rather than getting a notification for every minor change.

---

Concepts & Usage

This library introduces two core concepts for inter-process communication: host_variable and host_function. Just as a global variable or function can be accessed across different modules within a single process, these "host" objects can be accessed across different processes.

• host_variable
  A host_variable is a piece of shared memory identified by a unique name (a string). It's managed internally, and a host_variable object (a typedef of a struct pointer) acts as the handle to this shared memory.

• host_function
  A host_function is a function that can be called by another process. Arguments and results are transferred via named pipes. The process that defines the function runs a dispatcher thread that listens for incoming requests on these pipes.
  Functions are called asynchronously within a single dispatcher.
  Functions dispatched by different dispatchers run concurrently in different threads.

See apis.md for detailed api definition and python3-apis.md or python3-apis-cn.md for python interfaces.

---

Building the Library

This project uses CMake for its build system, making it easy to integrate into your existing project. Simply use add_subdirectory to include the main CMakeLists.txt file located in the root directory. Static linking is recommended for simplicity, but dynamic linking is also fully supported.

• BUILD_EXAMPLE
  To build the examples included with the library, set the CMake option BUILD_EXAMPLE to ON.
• BUILD_C
  Whether to build C interface into the .so file.
• BUILD_CPP
  Whether to build C++ interface into the .so file.

Example: Build with Examples

To build the library and the example programs, run the following commands from your project root:

mkdir build
cd build
cmake -DBUILD_EXAMPLE=ON ..
make

---

Python Installation

Install the Python binding after building the shared library as above:

cd robot_ipc
pip install .