Class AdaptiveComponent

Inheritance Relationships

Base Type

  • public rclcpp::Node

Class Documentation

class AdaptiveComponent : public rclcpp::Node

A composable container for Adaptive ROS 2 Node computations.

Select between FPGA, CPU or GPU at run-time.

Nodes using hardware acceleration are able to perform computations faster relying on FPGAs or GPUs, improving performance. Adaptive ROS 2 Nodes leverage hardware acceleration at run-time, allowing to select which computational resource the Node uses on-the-go, providing with a finer-grained control over the resources the computional graphs use in the underlying hardware.

AdaptiveComponent is a ROS 2 Node subclass programmed as a Component and including its own single threaded executor to build adaptive computations. Adaptive ROS 2 Nodes can then be built easily with it. Adaptive behavior is controlled through the “adaptive” ROS 2 parameter.

Public Types

enum Hardware

Values:

enumerator CPU
enumerator FPGA
enumerator GPU

Public Functions

inline RCLCPP_PUBLIC AdaptiveComponent(const rclcpp::NodeOptions &options)

Constructor.

Parameters:

options – [in] Options used to configure the executor.

inline RCLCPP_PUBLIC AdaptiveComponent(const rclcpp::NodeOptions &options, std::shared_ptr<rclcpp::Node> cpu_node, std::shared_ptr<rclcpp::Node> fpga_node, std::shared_ptr<rclcpp::Node> gpu_node, const int adaptive_value)

Constructor.

Parameters:

  • options – [in] Options used to configure the executor.

  • cpu_node – [in] std::shared_ptr<rclcpp::Node> Node for CPU computations.

  • fpga_node – [in] std::shared_ptr<rclcpp::Node> Node for FPGA computations.

  • fpga_node – [in] std::shared_ptr<rclcpp::Node> Node for GPU computations.

  • adaptive_value – [in] int initial value for “adaptive” ROS parameter. Defines what’s the initial compute substrate that’s used.

inline RCLCPP_PUBLIC AdaptiveComponent(const std::string &node_name, const rclcpp::NodeOptions &options, std::shared_ptr<rclcpp::Node> cpu_node = nullptr, std::shared_ptr<rclcpp::Node> fpga_node = nullptr, std::shared_ptr<rclcpp::Node> gpu_node = nullptr, const int adaptive_value = 0)

Default constructor.

Parameters:

  • node_name – [in] std::string Adaptive Node name.

  • options – [in] Options used to configure the executor.

  • cpu_node – [in] std::shared_ptr<rclcpp::Node> Node for CPU computations.

  • fpga_node – [in] std::shared_ptr<rclcpp::Node> Node for FPGA computations.

  • fpga_node – [in] std::shared_ptr<rclcpp::Node> Node for GPU computations.

  • adaptive_value – [in] int initial value for “adaptive” ROS parameter and defines what’s the initial compute substrate that’s used.

inline RCLCPP_PUBLIC void add_cpu (std::shared_ptr< rclcpp::Node > node_ptr)

Add CPU Node manually.

inline RCLCPP_PUBLIC void add_fpga (std::shared_ptr< rclcpp::Node > node_ptr)

Add FPGA Node manually.

inline RCLCPP_PUBLIC void add_gpu (std::shared_ptr< rclcpp::Node > node_ptr)

Add GPU Node manually.

Protected Functions

RCLCPP_PUBLIC void on_timer (void)

Timer callback at 1 second, used to update “adaptive” parameter.

RCLCPP_PUBLIC void spin (void)

Spin internal single threaded executor.

Protected Attributes

rclcpp::executors::SingleThreadedExecutor exec_

Internal AdaptiveComponent container’s executor.

std::shared_ptr<rclcpp::Node> cpu_node_

Computational Node for the CPU.

std::shared_ptr<rclcpp::Node> fpga_node_

Computational Node for the FPGA.

std::shared_ptr<rclcpp::Node> gpu_node_

Computational Node for the GPU.

rclcpp::TimerBase::SharedPtr timer_

AdaptiveCoponent internal timer.

int adaptive_value_

Mirrors the Node’s “adaptive” ROS 2 parameter.

std::shared_ptr<rclcpp::Node> compute_resources_[3]

Pointers to nodes, simplify interactions.