How can you optimize ROS node memory usage?

1 Use ROS topics wisely

ROS topics are the channels through which nodes exchange messages, making them essential for coordinating different functions and sensors of your robot. However, each topic has a buffer size that determines how many messages can be stored before they are discarded or overwritten, which can lead to wasting memory and bandwidth if not managed properly. To avoid this, you should use the rostopic list and rostopic info commands to check the number and size of your topics, as well as use the rostopic hz and rostopic bw commands to monitor the frequency and bandwidth. Additionally, you should use the queue_size parameter when creating publishers and subscribers to control the buffer size of your topics, as well as use the rostopic pub and rostopic echo commands to test and debug your topics.

2 Use dynamic memory allocation

Dynamic memory allocation is the process of requesting and releasing memory at runtime, depending on the needs of your program. This can help you save memory and avoid memory leaks, which are errors that occur when memory is not freed after it is no longer needed. However, dynamic memory allocation also has some drawbacks, such as fragmentation, overhead, and unpredictability. To use dynamic memory allocation effectively, you should consider using smart pointers such as std::shared_ptr or boost::shared_ptr to manage the lifetime and ownership of your dynamically allocated objects. Additionally, containers like std::vector or std::map can be used to store data in a flexible and efficient way. The new and delete operators should be used with caution, while malloc and free should be avoided in C++. Additionally, the valgrind tool can be used to detect and fix memory leaks and errors.

3 Use ROS services and actions

ROS services and actions are two alternative ways of communicating between nodes, besides topics. These methods are useful for tasks that require a specific outcome or progress report, such as moving a robot arm or navigating a map, and they have some advantages over topics in terms of memory usage. Services and actions use less memory, bandwidth, and CPU than topics, because they do not store messages in buffers, broadcast messages to multiple subscribers, or require polling or spinning. To use services and actions, you should use the rosservice list and rosservice info commands to check the available services and their types, as well as the rosaction list and rosaction info commands to check the available actions and their types. Additionally, use the rosservice call and rosaction send commands to invoke services and actions from the command line, or use the roscpp or rospy libraries to create and use service and action clients and servers in your code.

4 Use ROS parameters and config files

ROS parameters are global variables that can be accessed and modified by any node in the system, making them useful for storing and sharing configuration data, such as robot dimensions, sensor settings, or algorithm parameters. However, they can also have memory implications. For instance, the parameter server runs in the background and consumes memory, parameters are cached in each node that accesses them, and parameters are transmitted over the network which can consume bandwidth and affect latency. To use parameters efficiently, you should use the rosparam list and rosparam get commands to check existing parameters and their values; use the rosparam set and rosparam delete commands to modify or remove parameters from the parameter server; use the rosparam load and rosparam dump commands to load or save parameters from or to config files; and use the roscpp or rospy libraries to access and update parameters in your code.

6 Here’s what else to consider

This is a space to share examples, stories, or insights that don’t fit into any of the previous sections. What else would you like to add?