rbpodo 0.16.9

A client library for Rainbow Robotics' cobot RB-Series

RBPodo

[!WARNING] The API is currently under development and is subject to change.

This is a client library for Rainbow Robotics' cobots RB-Series. It is compatible with C++17.

You can find the documentation here.

---

Installation

To build rbpodo using CMake, just run

mkdir build
cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make

To install rbpodo for integrating your program, you just run

sudo make install

In your CMake project, you can include and link rbpodo

find_package(rbpodo REQUIRED)
target_link_libraries(<YOUR TARGET> rbpodo::rbpodo)

rbpodo is also available as a Python module. You can install it from PyPI via

pip install rbpodo

Or you can build and install Python module from source via

pip install .

Basic Example

You can find examples here.

C++

#include <iostream>
#include "rbpodo/rbpodo.hpp"

using namespace rb;

int main() {
  try {
    // Make connection
    podo::Cobot robot("10.0.2.7");
    podo::ResponseCollector rc;

    robot.set_operation_mode(rc, podo::OperationMode::Simulation);
    robot.set_speed_bar(rc, 0.5);

    robot.flush(rc);
    
    // Move robot in joint space
    robot.move_j(rc, {100, 0, 0, 0, 0, 0}, 200, 400);
    if (robot.wait_for_move_started(rc, 0.1).type() == podo::ReturnType::Success) {
      robot.wait_for_move_finished(rc);
    }
    // If there is any error during above process, throw exception error
    rc.error().throw_if_not_empty();
  } catch (const std::exception& e) {
    std::cerr << e.what() << std::endl;
    return 1;
  }
  return 0;
}

Python

import rbpodo as rb
import numpy as np

ROBOT_IP = "10.0.2.7"


def _main():
    try:
        robot = rb.Cobot(ROBOT_IP)
        rc = rb.ResponseCollector()

        robot.set_operation_mode(rc, rb.OperationMode.Simulation)
        robot.set_speed_bar(rc, 0.5)

        robot.flush(rc)

        robot.move_j(rc, np.array([100, 0, 0, 0, 0, 0]), 200, 400)
        if robot.wait_for_move_started(rc, 0.1).type() == rb.ReturnType.Success:
            robot.wait_for_move_finished(rc)
        rc.error().throw_if_not_empty()
    except Exception as e:
        print(e)
    finally:
        pass


if __name__ == "__main__":
    _main()

Joint Blending Move

blending_value = [0.01, 5.0, 20.0, 50.0]
q = []
for bv in blending_value:
    robot.move_jb2_clear(rc)
    robot.move_jb2_add(rc, np.array([90, 0, 0, 0, 0, 0]), 100, 100, bv)
    robot.move_jb2_add(rc, np.array([0, 0, 0, 0, 0, 0]), 100, 100, bv)
    robot.move_jb2_add(rc, np.array([90, 0, 0, 0, 0, 0]), 100, 100, bv)
    robot.move_jb2_add(rc, np.array([0, 0, 0, 0, 0, 0]), 100, 100, bv)
    robot.move_jb2_add(rc, np.array([90, 0, 0, 0, 0, 0]), 100, 100, bv)

    robot.flush(rc)
    robot.move_jb2_run(rc)

    data = []
    if robot.wait_for_move_started(rc, 0.5).type() == rb.ReturnType.Success:
        while robot.wait_for_move_finished(rc, 0.).type() == rb.ReturnType.Timeout:
            data.append(data_channel.request_data().sdata.jnt_ref)
            time.sleep(0.01)
        q.append(np.squeeze(np.array(data)[:, 0]))
q = np.vstack([np.hstack((e, np.tile(e[-1], max([e.shape[0] for e in q]) - e.shape[0]))) for e in q])

You can plot q via plt.plot(np.arange(0, q.shape[1]) * 0.01, np.transpose(q))

Wait functions

We provide wait functions for convenient external control. However, to achieve the intended behavior, it is necessary to understand the mechanism and use the wait functions accurately. Processes and the control box exchange commands and data through port 5000. After an external process sends a command, the control box sends back an ACK message indicating that the command has been executed (as shown in steps 1 and 2 in the dialog below). Afterward, whenever an event occurs (such as the start or end of move), the corresponding event message is delivered to all connected processes. The wait function operates by parsing this event message to check if a certain condition has been met.

If only one process is connected to the control box, this is generally not a problem. However, if multiple processes (or multiple computers) are connected, it may not work as expected. For example, in the illustration below, if while the wait_for_move_finished function is checking in process 1, some other process triggers an error unrelated to move ( such as trying to read a nonexistent variable), the control box will send an error message to all connected processes. In such cases, the wait function (if return_on_error is true) will assume an error has occurred and immediately return.

To solve this problem, one can either repeatedly check whether an error is related to movement or check whether the robot's state is idle. (or you can check though port 5001 data channel.)

while robot.get_robot_state(rc)[1] == rb.RobotState.Moving:
    time.sleep(1.0e-3)

Additionally, due to this mechanism, before using wait function, it is necessary to flush all messages in the buffer (a storage for messages from the control box that have not yet been processed) to avoid the program mistaking the result of a previous move (or command) for the result of the current command.

Advanced Topic

[!WARNING] This is experimental feature. Be careful when you use this.

Realtime script

rt_script() allows for the direct integration of custom scripts into the real-time control loop executed within the control box of robotic arm systems. By enabling computation to be carried out locally in the control box, it significantly reduces communication latency associated with the updating control output to robot arm. For instance, variables related to the feedback loop—such as joint positions and electrical current measurements—can be accessed directly in the control box.

The following is the part of example.

robot.eval(rc, "var count = 0");

robot.rt_script_onoff(rc, true);
robot.rt_script(rc, "count += 1");

for ( ... ) {
  std::string count_str;
  robot.print_variable(rc, "count", count_str);
  ...
}

robot.eval(rc, "var count = 0");