This page is part of the documentation of auKsys/5

WARNING auKsys/5 is still under development, and subject to changes. This documentation has not been fully updated with changes since auKsys/4.

The computation server is a ROS node that can be used to run computations dyamically, and in particular for data stored in a kDB database.

computation_server node

computation_server is the main node for running computation. After starting, it offers the following services:

  • check_status (ros_pralin_interfaces/srv/CheckStatus) it is used to check the state of a computation.
  • pause_processing (ros_pralin_interfaces/srv/ControlProcessing) it is used to pause a computation.
  • resume_processing (ros_pralin_interfaces/srv/ControlProcessing) it is used to resume a computation, after it has been paused.
  • start_processing (ros_pralin_interfaces/srv/StartProcessing) it is used to start a new computation.
  • stop_processing (ros_pralin_interfaces/srv/ControlProcessing) it is used to stop a computation.

Note that pause_processing and stop_processing can only interrupt a computation between two algorithms call. So if a computation is running for a long time in an algorithm, it won’t be stopped. check_status, pause_processing, resume_processing and stop_processing takes as an argument the UUID of the computation, the UUID was either given as an argument during start_processing, or returned in the response.

start_processing

The start_processing service use the following arguments:

  • definition (string): a YAML definition of the processing.
  • uuid (string): uuid to identify the processing (optional, if not given, a UUID will be generated and returned).
  • name_space (string): namespace to use for the topics created by the inputs/outputs. If not given, the topics are created in a namespace generated from the UUID.
  • parameters (string): a YAML string defining the parameters used by the computation.
  • repeat (bool): if true, the computation repeat on inputs, if false, it will ends once it has completed once.
  • inputs_triggers (string[]): if empty, any input will trigger a computation, if set, only the specified inputs will
  • inputs_mappings (Mapping[]): remap input topics
  • outputs_mappings (Mapping[]): remap output topics

If the composition defined in definition has inputs or outputs, the computation_server will attempt to create topics for those inputs and outputs. This require for the type of the input to be supported, otherwise, it will trigger an error. The topic are created using the name_space, or if no namespace is specified, using the uuid. Usually, in ROS, topics can be remapped when the node is started, but this is not possible with the computation server. This can be solved using the inputs_mappings and output_mappings. Mapping has the following fields:

  • connector_name (string): name of the input/output, as defined in the definition.
  • topic_name (string): name of the topic to re-map to. If the name starts with a /, it will be an absolute topic, otherwise it will be relative to the namespace of the server. `

run_computation

run_computation is a program that can be used to start a new computation from the command line.

It has the following usage:

run_computation [options] pralin_compose.yaml

Where pralin_compose.yaml is a filename with a composition definition. Alternatively, the composition can be specified in the terminal when using the --stdin option. The list of options are:

  • --help shows the help message.
  • --stdin read the composition definition from the terminal, instead of a file.
  • --repeat repeat the calculation until it is terminated by a service call.
  • --not-waiting by default, the script wait until the calculation is finished, used this option if you want the script to terminate as soon as the calculation is started.
  • --namespace ns define the namespace used by the topics created by the computation, it default to the UUID of the calculation
  • --parameters params a YAML string with the parameters used by the computation.

Example of use of this program can be found in the ROS Computation Server tutorial.