【附带源码】机械臂MoveIt2极简教程（二）、move_group交互

押波张飞

已于 2024-06-18 11:30:01 修改

阅读量1.2k

点赞数 21

分类专栏： MoveIt2极简教程文章标签： moveit2 moveit ros 机器人

于 2024-06-05 17:59:57 首次发布

本文链接：https://blog.csdn.net/ryanji/article/details/139472580

版权

MoveIt2极简教程专栏收录该内容

6 篇文章 3 订阅

订阅专栏

系列文章目录

【附带源码】机械臂MoveIt2极简教程（一）、moveit2安装
 【附带源码】机械臂MoveIt2极简教程（二）、move_group交互
 【附带源码】机械臂MoveIt2极简教程（三）、URDF/SRDF介绍
 【附带源码】机械臂MoveIt2极简教程（四）、第一个入门demo
【附带源码】机械臂MoveIt2极简教程（五）、第二个demo - rviz可视化

完成上一节的MoveIt2安装之后，大家也进行了demo演示，对MoveIt2有了初步了解。这一节通过C++包装的软件接口，实现机械臂完成更有意思的运动路径。

在这里插入图片描述

一、MoveGroup是什么？

在MoveIt中，最简单的交互接口就是通过调用MoveGroupInterface类来实现的。它为大多数机械臂操作提供了最简单的使用功能，比如在设置关节位置点或者目标点后，想要用机械臂抓取东西；在现实环境中增加障碍物，让机械臂能绕障运动等等。这些都可以通过MoveGroup类中实现的ros node节点来完成。

二、实现效果

首先咱们本节的实现效果，心中有了效果图，再回头看代码，更加能够理解其中含义，也不会枯燥。

1. 加载机械臂

打开两个terminal
第一个terminal输入

ros2 launch moveit2_tutorials move_group.launch.py

另一个terminal输入

ros2 launch moveit2_tutorials move_group_interface_tutorial.launch.py

之后出现rviz界面，如下图所示
在这里插入图片描述

2. 轨迹规划、导航

想要让机械臂运动起来，有两种方法：

使用Key Tool工具：在rviz中的第二行工具栏点击＋号，添加Key Tool工具；点击Key Tool，键盘按0即可让机械臂执行绕障运动等等动作。可以连续点击。
使用RvizVisualToolsGui面板：点击next按钮同样可以实现机械臂运动。可以连续点击。

3. 运动过程罗列

一共有10个运动过程：

移动机械臂末端移动到它前方目标点；
移动机械臂末端移动到它侧边目标点；
移动机械臂末端移动到后方的目标点，同时保持末端执行水平；
移动机械臂末端沿着笛卡尔路径运动（下、右、斜左上）；
移动机械臂末端移动到前方某一个目标点（中间无障碍物）；
添加一个隔板障碍物；
机械臂越过隔板，避障到达目标点；
机械臂末端执行器下方携带一个紫色圆柱体的物体（模拟运输物体）；
机械臂带着圆柱体从侧身越过隔板障碍物，到达目标点（模拟运输物体过程的避障）；
移除圆柱体；
移除隔板障碍物；

三、几个常见术语

在看代码之前，我们有必要进行常见术语的简要描述。既然这个系列是姨妈级呵护的教程，那么术语的解释也要通俗易懂。

末端执行器：就是夹爪，哈哈哈哈
轨迹规划
可能你会说，轨迹规划我早就知道，不就是在当前位置和目标位置之间撒一些路径轨迹点吗？事实确实如此，但是机械臂的轨迹规划与机器人移动小车的轨迹规划有区别。

机械臂的轨迹规划是要把末端执行器位姿分解到每个关节的位姿变换。可以这样理解，以你的手臂举例，如果你想要一拳打在你前方的人脸上，拳头怎么到达是由你肩部关节、肘部关节、手腕关节联合运动的合力。拳头就是你的末端执行器。

关节空间规划与笛卡尔空间规划

关节空间规划：通过确定各个关节的转角，并组合成一个序列，才能完成末端执行器的轨迹规划。优点：计算简单，只要计算每个关节转角即可。缺点：末端控制不够精确。

笛卡尔空间规划：通过末端执行器的位姿，在起点和终点之间撒轨迹点，对每个轨迹点利用逆运动学分解得到每个关节的转角。最后会得到一系列的关节转角序列。优点：精确控制，避障。缺点：逆运动学计算复杂，同时轨迹要插值，增加计算量。

笛卡尔空间规划其实和移动机器人的基于运动的路径规划算法大致相似（在不考虑机械臂关节转角的前提下），都是先来一个全局路径，在撒点，根据每个点进行一个局部路径规划。现在流行的特斯拉基于大模型的端到端轨迹规划导航则是另一种方法，没有全局路径和局部路径的分类。这是后话。

四、简要解读代码

代码位于moveit2_tutorials包的/doc/move_group_interface/src/move_group_interface_tutorial.cpp中。github源码链接

1. 基本设置

这些是ros2相关的初始化，与moveit2无关:

rclcpp::init(argc, argv);
rclcpp::NodeOptions node_options;
node_options.automatically_declare_parameters_from_overrides(true);
auto move_group_node = rclcpp::Node::make_shared("move_group_interface_tutorial", node_options);

moveit会在一系列的关节上进行操作，统称planning groups，存储在一个JointModelGroup的实例中。

static const std::string PLANNING_GROUP = "panda_arm";

一旦你设置了planning group的名字，那么MoveGroupInterface类就会进行绑定，之后你所有的控制和规划都与它相关。

moveit::planning_interface::MoveGroupInterface move_group(move_group_node, PLANNING_GROUP);

PlanningSceneInterface类则是用来添加和移除障碍物。

moveit::planning_interface::PlanningSceneInterface planning_scene_interface;

JointModelGroup的指针会频繁地被用来指向planning group，这样可以大大提高执行机构的性能。

const moveit::core::JointModelGroup* joint_model_group =
    move_group.getCurrentState()->getJointModelGroup(PLANNING_GROUP);

2. 可视化

rviz提供了很多markers，这个例子中使用了文本、圆柱体、球体。

namespace rvt = rviz_visual_tools;
moveit_visual_tools::MoveItVisualTools visual_tools(move_group_node, "panda_link0", "move_group_tutorial",
                                                    move_group.getRobotModel());

visual_tools.deleteAllMarkers();

/* Remote control is an introspection tool that allows users to step through a high level script */
/* via buttons and keyboard shortcuts in RViz */
visual_tools.loadRemoteControl();
Eigen::Isometry3d text_pose = Eigen::Isometry3d::Identity();
text_pose.translation().z() = 1.0;
visual_tools.publishText(text_pose, "MoveGroupInterface_Demo", rvt::WHITE, rvt::XLARGE);

为了提高发给rviz的大数据量的传输速率，moveit可视化工具进行了优化传输：

visual_tools.trigger();

3. 获取基本信息

打印机械臂参考系的名字

RCLCPP_INFO(LOGGER, "Planning frame: %s", move_group.getPlanningFrame().c_str());

打印规划组planning group的末端执行器的名字

RCLCPP_INFO(LOGGER, "End effector link: %s", move_group.getEndEffectorLink().c_str());

获取机械臂所有的规划组名字

RCLCPP_INFO(LOGGER, "Available Planning Groups:");
std::copy(move_group.getJointModelGroupNames().begin(), move_group.getJointModelGroupNames().end(),
          std::ostream_iterator<std::string>(std::cout, ", "));

4. 运动过程解释

单点轨迹规划

为末端执行器规划一个合适的位姿，

geometry_msgs::msg::Pose target_pose1;
target_pose1.orientation.w = 1.0;
target_pose1.position.x = 0.28;
target_pose1.position.y = -0.2;
target_pose1.position.z = 0.5;
move_group.setPoseTarget(target_pose1);

调用规划器来生成轨迹，并可视化显示。注意现在仅仅是规划，没有进行导航移动

moveit::planning_interface::MoveGroupInterface::Plan my_plan;

bool success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);

RCLCPP_INFO(LOGGER, "Visualizing plan 1 (pose goal) %s", success ? "" : "FAILED");

可视化轨迹路径

RCLCPP_INFO(LOGGER, "Visualizing plan 1 as trajectory line");
visual_tools.publishAxisLabeled(target_pose1, "pose1");
visual_tools.publishText(text_pose, "Pose_Goal", rvt::WHITE, rvt::XLARGE);
visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
visual_tools.trigger();
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

单点导航

现实中的本体机械臂在规划完路径之后直接执行move_group.move()即可，但是仿真不能这么用。因为这个函数是个阻塞函数，必须得到控制器的成功返回指令，否则一直堵塞。

关节空间规划

一开始，创建一个指针指向当前机械臂的位姿状态。机械臂状态包括所有当前的位姿、速度、加速度等

moveit::core::RobotStatePtr current_state = move_group.getCurrentState(10);

之后，获得当前关节组的所有关节值

std::vector<double> joint_group_positions;
current_state->copyJointGroupPositions(joint_model_group, joint_group_positions);

修改其中一个关节值，规划新的关节空间路径，可视化路径

joint_group_positions[0] = -1.0;  // radians
bool within_bounds = move_group.setJointValueTarget(joint_group_positions);
if (!within_bounds)
{
  RCLCPP_WARN(LOGGER, "Target joint position(s) were outside of limits, but we will plan and clamp to the limits ");
}

把速度和加速度都调低一点。默认值是0.1（最大值的10%）
这些配置都在moveit_config中的joint_limits.yaml中

move_group.setMaxVelocityScalingFactor(0.05);
move_group.setMaxAccelerationScalingFactor(0.05);

success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
RCLCPP_INFO(LOGGER, "Visualizing plan 2 (joint space goal) %s", success ? "" : "FAILED");

可视化：

visual_tools.deleteAllMarkers();
visual_tools.publishText(text_pose, "Joint_Space_Goal", rvt::WHITE, rvt::XLARGE);
visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
visual_tools.trigger();
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

路径约束

只要对某一个目标点的位姿添加路径约束即可，然后跟move_group绑定。定向路径约束的轨迹规划通常是在笛卡尔空间中进行采样。

moveit_msgs::msg::OrientationConstraint ocm;
ocm.link_name = "panda_link7";
ocm.header.frame_id = "panda_link0";
ocm.orientation.w = 1.0;
ocm.absolute_x_axis_tolerance = 0.1;
ocm.absolute_y_axis_tolerance = 0.1;
ocm.absolute_z_axis_tolerance = 0.1;
ocm.weight = 1.0;


moveit_msgs::msg::Constraints test_constraints;
test_constraints.orientation_constraints.push_back(ocm);
move_group.setPathConstraints(test_constraints);

强制关节空间规划

MoveIt会根据实际问题自动选择关节空间规划还是笛卡尔空间规划。如果你想强制使用关节空间规划，那么需要在ompl_planning.yaml中修改参数enforce_joint_model_state_space，设为true。
设置开始状态

moveit::core::RobotState start_state(*move_group.getCurrentState());
geometry_msgs::msg::Pose start_pose2;
start_pose2.orientation.w = 1.0;
start_pose2.position.x = 0.55;
start_pose2.position.y = -0.05;
start_pose2.position.z = 0.8;
start_state.setFromIK(joint_model_group, start_pose2);
move_group.setStartState(start_state);

早期位姿规划

move_group.setPoseTarget(target_pose1);

带有约束的路径规划最好速度慢一点，因为每一个采样必须调用逆运动解的求解器。增加规划时间，这样留给规划器的时间会多一点，那么规划成功的概率就大一点。

move_group.setPlanningTime(10.0);

success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
RCLCPP_INFO(LOGGER, "Visualizing plan 3 (constraints) %s", success ? "" : "FAILED");

可视化路径

visual_tools.deleteAllMarkers();
visual_tools.publishAxisLabeled(start_pose2, "start");
visual_tools.publishAxisLabeled(target_pose1, "goal");
visual_tools.publishText(text_pose, "Constrained_Goal", rvt::WHITE, rvt::XLARGE);
visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
visual_tools.trigger();
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

最后记得清除路径约束

move_group.clearPathConstraints();

笛卡尔空间轨迹规划

通过给末端执行器设置一系列的waypoints，最后就可以生成笛卡尔路径。初始姿态不需要加到waypoints列表中，但是可视化的时候可以加进来，方便理解。

std::vector<geometry_msgs::msg::Pose> waypoints;
waypoints.push_back(start_pose2);

geometry_msgs::msg::Pose target_pose3 = start_pose2;

target_pose3.position.z -= 0.2;
waypoints.push_back(target_pose3);  // down

target_pose3.position.y -= 0.2;
waypoints.push_back(target_pose3);  // right

target_pose3.position.z += 0.2;
target_pose3.position.y += 0.2;
target_pose3.position.x -= 0.2;
waypoints.push_back(target_pose3);  // up and left

设置插值分辨率为1cm，因此笛卡尔移动最大步长是0.01m。jump 阈值要设置为0.0，目的是不让它有跳跃点产生。但是在现实中，关闭jump可能会造成未知的运动，因此有可能是个安全隐患。

moveit_msgs::msg::RobotTrajectory trajectory;
const double jump_threshold = 0.0;
const double eef_step = 0.01;
double fraction = move_group.computeCartesianPath(waypoints, eef_step, jump_threshold, trajectory);
RCLCPP_INFO(LOGGER, "Visualizing plan 4 (Cartesian path) (%.2f%% achieved)", fraction * 100.0);

可视化

visual_tools.deleteAllMarkers();
visual_tools.publishText(text_pose, "Cartesian_Path", rvt::WHITE, rvt::XLARGE);
visual_tools.publishPath(waypoints, rvt::LIME_GREEN, rvt::SMALL);
for (std::size_t i = 0; i < waypoints.size(); ++i)
  visual_tools.publishAxisLabeled(waypoints[i], "pt" + std::to_string(i), rvt::SMALL);
visual_tools.trigger();
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

笛卡尔运动一定要慢，特别是靠近物体时。笛卡尔规划当前不能通过maxVelocityScalingFactor来设置，需要你手动计算轨迹行进的时间。

环境中添加障碍物

无障碍物规划

move_group.setStartState(*move_group.getCurrentState());
geometry_msgs::msg::Pose another_pose;
another_pose.orientation.w = 0;
another_pose.orientation.x = -1.0;
another_pose.position.x = 0.7;
another_pose.position.y = 0.0;
another_pose.position.z = 0.59;
move_group.setPoseTarget(another_pose);

success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
RCLCPP_INFO(LOGGER, "Visualizing plan 5 (with no obstacles) %s", success ? "" : "FAILED");

visual_tools.deleteAllMarkers();
visual_tools.publishText(text_pose, "Clear_Goal", rvt::WHITE, rvt::XLARGE);
visual_tools.publishAxisLabeled(another_pose, "goal");
visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
visual_tools.trigger();
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

no objects plan
添加一个障碍物，让机械臂避障

moveit_msgs::msg::CollisionObject collision_object;
collision_object.header.frame_id = move_group.getPlanningFrame();
collision_object.id = "box1";
shape_msgs::msg::SolidPrimitive primitive;
primitive.type = primitive.BOX;
primitive.dimensions.resize(3);
primitive.dimensions[primitive.BOX_X] = 0.1;
primitive.dimensions[primitive.BOX_Y] = 1.5;
primitive.dimensions[primitive.BOX_Z] = 0.5;

定义障碍物的位姿

geometry_msgs::msg::Pose box_pose;
box_pose.orientation.w = 1.0;
box_pose.position.x = 0.48;
box_pose.position.y = 0.0;
box_pose.position.z = 0.25;

collision_object.primitives.push_back(primitive);
collision_object.primitive_poses.push_back(box_pose);
collision_object.operation = collision_object.ADD;

std::vector<moveit_msgs::msg::CollisionObject> collision_objects;
collision_objects.push_back(collision_object);

RCLCPP_INFO(LOGGER, "Add an object into the world");
planning_scene_interface.addCollisionObjects(collision_objects);

visual_tools.publishText(text_pose, "Add_object", rvt::WHITE, rvt::XLARGE);
visual_tools.trigger();
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to once the collision object appears in RViz");

当我们规划一个轨迹时，它就会开始避障

success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
RCLCPP_INFO(LOGGER, "Visualizing plan 6 (pose goal move around cuboid) %s", success ? "" : "FAILED");
visual_tools.publishText(text_pose, "Obstacle_Goal", rvt::WHITE, rvt::XLARGE);
visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
visual_tools.trigger();
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window once the plan is complete");

obstacle plan

携带物体的末端执行器

为什么要有这个demo？为了增加避障难度，因为这时候你要考虑你爪子上的物体不跟前面的障碍物发生碰撞。通常在工业领域，你的机械臂都是要带物操作的，你肯定不会让你携带的物体被撞碎吧。

moveit_msgs::msg::CollisionObject object_to_attach;
object_to_attach.id = "cylinder1";

shape_msgs::msg::SolidPrimitive cylinder_primitive;
cylinder_primitive.type = primitive.CYLINDER;
cylinder_primitive.dimensions.resize(2);
cylinder_primitive.dimensions[primitive.CYLINDER_HEIGHT] = 0.20;
cylinder_primitive.dimensions[primitive.CYLINDER_RADIUS] = 0.04;

定义物体的位姿和frame，这样可以出现在夹爪下方

object_to_attach.header.frame_id = move_group.getEndEffectorLink();
geometry_msgs::msg::Pose grab_pose;
grab_pose.orientation.w = 1.0;
grab_pose.position.z = 0.2;

添加障碍物

object_to_attach.primitives.push_back(cylinder_primitive);
object_to_attach.primitive_poses.push_back(grab_pose);
object_to_attach.operation = object_to_attach.ADD;
planning_scene_interface.applyCollisionObject(object_to_attach);

给夹爪加个物体。

RCLCPP_INFO(LOGGER, "Attach the object to the robot");
std::vector<std::string> touch_links;
touch_links.push_back("panda_rightfinger");
touch_links.push_back("panda_leftfinger");
move_group.attachObject(object_to_attach.id, "panda_hand", touch_links);

visual_tools.publishText(text_pose, "Object_attached_to_robot", rvt::WHITE, rvt::XLARGE);
visual_tools.trigger();

/* Wait for MoveGroup to receive and process the attached collision object message */
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window once the new object is attached to the robot");

有物在手，下雨不愁。。。

move_group.setStartStateToCurrentState();
success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
RCLCPP_INFO(LOGGER, "Visualizing plan 7 (move around cuboid with cylinder) %s", success ? "" : "FAILED");
visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
visual_tools.trigger();
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window once the plan is complete");

在这里插入图片描述

移除所有障碍物和夹爪下的物体

首先去除夹爪抓的物体

RCLCPP_INFO(LOGGER, "Detach the object from the robot");
move_group.detachObject(object_to_attach.id);

rviz中显示状态

visual_tools.deleteAllMarkers();
visual_tools.publishText(text_pose, "Object_detached_from_robot", rvt::WHITE, rvt::XLARGE);
visual_tools.trigger();

/* Wait for MoveGroup to receive and process the attached collision object message */
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window once the new object is detached from the robot");

移除障碍物

RCLCPP_INFO(LOGGER, "Remove the objects from the world");
std::vector<std::string> object_ids;
object_ids.push_back(collision_object.id);
object_ids.push_back(object_to_attach.id);
planning_scene_interface.removeCollisionObjects(object_ids);

rviz显示状态

visual_tools.publishText(text_pose, "Objects_removed", rvt::WHITE, rvt::XLARGE);
visual_tools.trigger();

/* Wait for MoveGroup to receive and process the attached collision object message */
visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to once the collision object disappears");

五、完整代码

/*********************************************************************
 * Software License Agreement (BSD License)
 *
 *  Copyright (c) 2013, SRI International
 *  All rights reserved.
 *
 *  Redistribution and use in source and binary forms, with or without
 *  modification, are permitted provided that the following conditions
 *  are met:
 *
 *   * Redistributions of source code must retain the above copyright
 *     notice, this list of conditions and the following disclaimer.
 *   * Redistributions in binary form must reproduce the above
 *     copyright notice, this list of conditions and the following
 *     disclaimer in the documentation and/or other materials provided
 *     with the distribution.
 *   * Neither the name of SRI International nor the names of its
 *     contributors may be used to endorse or promote products derived
 *     from this software without specific prior written permission.
 *
 *  THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
 *  "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
 *  LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
 *  FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
 *  COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
 *  INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
 *  BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
 *  LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
 *  CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
 *  LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN
 *  ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
 *  POSSIBILITY OF SUCH DAMAGE.
 *********************************************************************/

/* Author: Sachin Chitta, Dave Coleman, Mike Lautman */

#include <moveit/move_group_interface/move_group_interface.h>
#include <moveit/planning_scene_interface/planning_scene_interface.h>

#include <moveit_msgs/msg/display_robot_state.hpp>
#include <moveit_msgs/msg/display_trajectory.hpp>

#include <moveit_msgs/msg/attached_collision_object.hpp>
#include <moveit_msgs/msg/collision_object.hpp>

#include <moveit_visual_tools/moveit_visual_tools.h>

// All source files that use ROS logging should define a file-specific
// static const rclcpp::Logger named LOGGER, located at the top of the file
// and inside the namespace with the narrowest scope (if there is one)
static const rclcpp::Logger LOGGER = rclcpp::get_logger("move_group_demo");

int main(int argc, char** argv)
{
  rclcpp::init(argc, argv);
  rclcpp::NodeOptions node_options;
  node_options.automatically_declare_parameters_from_overrides(true);
  auto move_group_node = rclcpp::Node::make_shared("move_group_interface_tutorial", node_options);

  // We spin up a SingleThreadedExecutor for the current state monitor to get information
  // about the robot's state.
  rclcpp::executors::SingleThreadedExecutor executor;
  executor.add_node(move_group_node);
  std::thread([&executor]() { executor.spin(); }).detach();

  // BEGIN_TUTORIAL
  //
  // Setup
  // ^^^^^
  //
  // MoveIt operates on sets of joints called "planning groups" and stores them in an object called
  // the ``JointModelGroup``. Throughout MoveIt, the terms "planning group" and "joint model group"
  // are used interchangeably.
  static const std::string PLANNING_GROUP = "panda_arm";

  // The
  // :moveit_codedir:`MoveGroupInterface<moveit_ros/planning_interface/move_group_interface/include/moveit/move_group_interface/move_group_interface.h>`
  // class can be easily set up using just the name of the planning group you would like to control and plan for.
  moveit::planning_interface::MoveGroupInterface move_group(move_group_node, PLANNING_GROUP);

  // We will use the
  // :moveit_codedir:`PlanningSceneInterface<moveit_ros/planning_interface/planning_scene_interface/include/moveit/planning_scene_interface/planning_scene_interface.h>`
  // class to add and remove collision objects in our "virtual world" scene
  moveit::planning_interface::PlanningSceneInterface planning_scene_interface;

  // Raw pointers are frequently used to refer to the planning group for improved performance.
  const moveit::core::JointModelGroup* joint_model_group =
      move_group.getCurrentState()->getJointModelGroup(PLANNING_GROUP);

  // Visualization
  // ^^^^^^^^^^^^^
  namespace rvt = rviz_visual_tools;
  moveit_visual_tools::MoveItVisualTools visual_tools(move_group_node, "panda_link0", "move_group_tutorial",
                                                      move_group.getRobotModel());

  visual_tools.deleteAllMarkers();

  /* Remote control is an introspection tool that allows users to step through a high level script */
  /* via buttons and keyboard shortcuts in RViz */
  visual_tools.loadRemoteControl();

  // RViz provides many types of markers, in this demo we will use text, cylinders, and spheres
  Eigen::Isometry3d text_pose = Eigen::Isometry3d::Identity();
  text_pose.translation().z() = 1.0;
  visual_tools.publishText(text_pose, "MoveGroupInterface_Demo", rvt::WHITE, rvt::XLARGE);

  // Batch publishing is used to reduce the number of messages being sent to RViz for large visualizations
  visual_tools.trigger();

  // Getting Basic Information
  // ^^^^^^^^^^^^^^^^^^^^^^^^^
  //
  // We can print the name of the reference frame for this robot.
  RCLCPP_INFO(LOGGER, "Planning frame: %s", move_group.getPlanningFrame().c_str());

  // We can also print the name of the end-effector link for this group.
  RCLCPP_INFO(LOGGER, "End effector link: %s", move_group.getEndEffectorLink().c_str());

  // We can get a list of all the groups in the robot:
  RCLCPP_INFO(LOGGER, "Available Planning Groups:");
  std::copy(move_group.getJointModelGroupNames().begin(), move_group.getJointModelGroupNames().end(),
            std::ostream_iterator<std::string>(std::cout, ", "));

  // Start the demo
  // ^^^^^^^^^^^^^^^^^^^^^^^^^
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to start the demo");

  // .. _move_group_interface-planning-to-pose-goal:
  //
  // Planning to a Pose goal
  // ^^^^^^^^^^^^^^^^^^^^^^^
  // We can plan a motion for this group to a desired pose for the
  // end-effector.
  geometry_msgs::msg::Pose target_pose1;
  target_pose1.orientation.w = 1.0;
  target_pose1.position.x = 0.28;
  target_pose1.position.y = -0.2;
  target_pose1.position.z = 0.5;
  move_group.setPoseTarget(target_pose1);

  // Now, we call the planner to compute the plan and visualize it.
  // Note that we are just planning, not asking move_group
  // to actually move the robot.
  moveit::planning_interface::MoveGroupInterface::Plan my_plan;

  bool success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);

  RCLCPP_INFO(LOGGER, "Visualizing plan 1 (pose goal) %s", success ? "" : "FAILED");

  // Visualizing plans
  // ^^^^^^^^^^^^^^^^^
  // We can also visualize the plan as a line with markers in RViz.
  RCLCPP_INFO(LOGGER, "Visualizing plan 1 as trajectory line");
  visual_tools.publishAxisLabeled(target_pose1, "pose1");
  visual_tools.publishText(text_pose, "Pose_Goal", rvt::WHITE, rvt::XLARGE);
  visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
  visual_tools.trigger();
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

  // Moving to a pose goal
  // ^^^^^^^^^^^^^^^^^^^^^
  //
  // Moving to a pose goal is similar to the step above
  // except we now use the ``move()`` function. Note that
  // the pose goal we had set earlier is still active
  // and so the robot will try to move to that goal. We will
  // not use that function in this tutorial since it is
  // a blocking function and requires a controller to be active
  // and report success on execution of a trajectory.

  /* Uncomment below line when working with a real robot */
  /* move_group.move(); */

  // Planning to a joint-space goal
  // ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  //
  // Let's set a joint space goal and move towards it.  This will replace the
  // pose target we set above.
  //
  // To start, we'll create an pointer that references the current robot's state.
  // RobotState is the object that contains all the current position/velocity/acceleration data.
  moveit::core::RobotStatePtr current_state = move_group.getCurrentState(10);
  //
  // Next get the current set of joint values for the group.
  std::vector<double> joint_group_positions;
  current_state->copyJointGroupPositions(joint_model_group, joint_group_positions);

  // Now, let's modify one of the joints, plan to the new joint space goal, and visualize the plan.
  joint_group_positions[0] = -1.0;  // radians
  bool within_bounds = move_group.setJointValueTarget(joint_group_positions);
  if (!within_bounds)
  {
    RCLCPP_WARN(LOGGER, "Target joint position(s) were outside of limits, but we will plan and clamp to the limits ");
  }

  // We lower the allowed maximum velocity and acceleration to 5% of their maximum.
  // The default values are 10% (0.1).
  // Set your preferred defaults in the joint_limits.yaml file of your robot's moveit_config
  // or set explicit factors in your code if you need your robot to move faster.
  move_group.setMaxVelocityScalingFactor(0.05);
  move_group.setMaxAccelerationScalingFactor(0.05);

  success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
  RCLCPP_INFO(LOGGER, "Visualizing plan 2 (joint space goal) %s", success ? "" : "FAILED");

  // Visualize the plan in RViz:
  visual_tools.deleteAllMarkers();
  visual_tools.publishText(text_pose, "Joint_Space_Goal", rvt::WHITE, rvt::XLARGE);
  visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
  visual_tools.trigger();
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

  // Planning with Path Constraints
  // ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  //
  // Path constraints can easily be specified for a link on the robot.
  // Let's specify a path constraint and a pose goal for our group.
  // First define the path constraint.
  moveit_msgs::msg::OrientationConstraint ocm;
  ocm.link_name = "panda_link7";
  ocm.header.frame_id = "panda_link0";
  ocm.orientation.w = 1.0;
  ocm.absolute_x_axis_tolerance = 0.1;
  ocm.absolute_y_axis_tolerance = 0.1;
  ocm.absolute_z_axis_tolerance = 0.1;
  ocm.weight = 1.0;

  // Now, set it as the path constraint for the group.
  moveit_msgs::msg::Constraints test_constraints;
  test_constraints.orientation_constraints.push_back(ocm);
  move_group.setPathConstraints(test_constraints);

  // Enforce Planning in Joint Space
  // ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  //
  // Depending on the planning problem MoveIt chooses between
  // ``joint space`` and ``cartesian space`` for problem representation.
  // Setting the group parameter ``enforce_joint_model_state_space:true`` in
  // the ompl_planning.yaml file enforces the use of ``joint space`` for all plans.
  //
  // By default, planning requests with orientation path constraints
  // are sampled in ``cartesian space`` so that invoking IK serves as a
  // generative sampler.
  //
  // By enforcing ``joint space``, the planning process will use rejection
  // sampling to find valid requests. Please note that this might
  // increase planning time considerably.
  //
  // We will reuse the old goal that we had and plan to it.
  // Note that this will only work if the current state already
  // satisfies the path constraints. So we need to set the start
  // state to a new pose.
  moveit::core::RobotState start_state(*move_group.getCurrentState());
  geometry_msgs::msg::Pose start_pose2;
  start_pose2.orientation.w = 1.0;
  start_pose2.position.x = 0.55;
  start_pose2.position.y = -0.05;
  start_pose2.position.z = 0.8;
  start_state.setFromIK(joint_model_group, start_pose2);
  move_group.setStartState(start_state);

  // Now, we will plan to the earlier pose target from the new
  // start state that we just created.
  move_group.setPoseTarget(target_pose1);

  // Planning with constraints can be slow because every sample must call an inverse kinematics solver.
  // Let's increase the planning time from the default 5 seconds to be sure the planner has enough time to succeed.
  move_group.setPlanningTime(10.0);

  success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
  RCLCPP_INFO(LOGGER, "Visualizing plan 3 (constraints) %s", success ? "" : "FAILED");

  // Visualize the plan in RViz:
  visual_tools.deleteAllMarkers();
  visual_tools.publishAxisLabeled(start_pose2, "start");
  visual_tools.publishAxisLabeled(target_pose1, "goal");
  visual_tools.publishText(text_pose, "Constrained_Goal", rvt::WHITE, rvt::XLARGE);
  visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
  visual_tools.trigger();
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

  // When done with the path constraint, be sure to clear it.
  move_group.clearPathConstraints();

  // Cartesian Paths
  // ^^^^^^^^^^^^^^^
  // You can plan a Cartesian path directly by specifying a list of waypoints
  // for the end-effector to go through. Note that we are starting
  // from the new start state above.  The initial pose (start state) does not
  // need to be added to the waypoint list but adding it can help with visualizations
  std::vector<geometry_msgs::msg::Pose> waypoints;
  waypoints.push_back(start_pose2);

  geometry_msgs::msg::Pose target_pose3 = start_pose2;

  target_pose3.position.z -= 0.2;
  waypoints.push_back(target_pose3);  // down

  target_pose3.position.y -= 0.2;
  waypoints.push_back(target_pose3);  // right

  target_pose3.position.z += 0.2;
  target_pose3.position.y += 0.2;
  target_pose3.position.x -= 0.2;
  waypoints.push_back(target_pose3);  // up and left

  // We want the Cartesian path to be interpolated at a resolution of 1 cm
  // which is why we will specify 0.01 as the max step in Cartesian
  // translation.  We will specify the jump threshold as 0.0, effectively disabling it.
  // Warning - disabling the jump threshold while operating real hardware can cause
  // large unpredictable motions of redundant joints and could be a safety issue
  moveit_msgs::msg::RobotTrajectory trajectory;
  const double jump_threshold = 0.0;
  const double eef_step = 0.01;
  double fraction = move_group.computeCartesianPath(waypoints, eef_step, jump_threshold, trajectory);
  RCLCPP_INFO(LOGGER, "Visualizing plan 4 (Cartesian path) (%.2f%% achieved)", fraction * 100.0);

  // Visualize the plan in RViz
  visual_tools.deleteAllMarkers();
  visual_tools.publishText(text_pose, "Cartesian_Path", rvt::WHITE, rvt::XLARGE);
  visual_tools.publishPath(waypoints, rvt::LIME_GREEN, rvt::SMALL);
  for (std::size_t i = 0; i < waypoints.size(); ++i)
    visual_tools.publishAxisLabeled(waypoints[i], "pt" + std::to_string(i), rvt::SMALL);
  visual_tools.trigger();
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

  // Cartesian motions should often be slow, e.g. when approaching objects. The speed of Cartesian
  // plans cannot currently be set through the maxVelocityScalingFactor, but requires you to time
  // the trajectory manually, as described `here <https://groups.google.com/forum/#!topic/moveit-users/MOoFxy2exT4>`_.
  // Pull requests are welcome.
  //
  // You can execute a trajectory like this.
  /* move_group.execute(trajectory); */

  // Adding objects to the environment
  // ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  //
  // First, let's plan to another simple goal with no objects in the way.
  move_group.setStartState(*move_group.getCurrentState());
  geometry_msgs::msg::Pose another_pose;
  another_pose.orientation.w = 0;
  another_pose.orientation.x = -1.0;
  another_pose.position.x = 0.7;
  another_pose.position.y = 0.0;
  another_pose.position.z = 0.59;
  move_group.setPoseTarget(another_pose);

  success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
  RCLCPP_INFO(LOGGER, "Visualizing plan 5 (with no obstacles) %s", success ? "" : "FAILED");

  visual_tools.deleteAllMarkers();
  visual_tools.publishText(text_pose, "Clear_Goal", rvt::WHITE, rvt::XLARGE);
  visual_tools.publishAxisLabeled(another_pose, "goal");
  visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
  visual_tools.trigger();
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to continue the demo");

  // The result may look like this:
  //
  // .. image:: ./move_group_interface_tutorial_clear_path.gif
  //    :alt: animation showing the arm moving relatively straight toward the goal
  //
  // Now, let's define a collision object ROS message for the robot to avoid.
  moveit_msgs::msg::CollisionObject collision_object;
  collision_object.header.frame_id = move_group.getPlanningFrame();

  // The id of the object is used to identify it.
  collision_object.id = "box1";

  // Define a box to add to the world.
  shape_msgs::msg::SolidPrimitive primitive;
  primitive.type = primitive.BOX;
  primitive.dimensions.resize(3);
  primitive.dimensions[primitive.BOX_X] = 0.1;
  primitive.dimensions[primitive.BOX_Y] = 1.5;
  primitive.dimensions[primitive.BOX_Z] = 0.5;

  // Define a pose for the box (specified relative to frame_id).
  geometry_msgs::msg::Pose box_pose;
  box_pose.orientation.w = 1.0;
  box_pose.position.x = 0.48;
  box_pose.position.y = 0.0;
  box_pose.position.z = 0.25;

  collision_object.primitives.push_back(primitive);
  collision_object.primitive_poses.push_back(box_pose);
  collision_object.operation = collision_object.ADD;

  std::vector<moveit_msgs::msg::CollisionObject> collision_objects;
  collision_objects.push_back(collision_object);

  // Now, let's add the collision object into the world
  // (using a vector that could contain additional objects)
  RCLCPP_INFO(LOGGER, "Add an object into the world");
  planning_scene_interface.addCollisionObjects(collision_objects);

  // Show text in RViz of status and wait for MoveGroup to receive and process the collision object message
  visual_tools.publishText(text_pose, "Add_object", rvt::WHITE, rvt::XLARGE);
  visual_tools.trigger();
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to once the collision object appears in RViz");

  // Now, when we plan a trajectory it will avoid the obstacle.
  success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
  RCLCPP_INFO(LOGGER, "Visualizing plan 6 (pose goal move around cuboid) %s", success ? "" : "FAILED");
  visual_tools.publishText(text_pose, "Obstacle_Goal", rvt::WHITE, rvt::XLARGE);
  visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
  visual_tools.trigger();
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window once the plan is complete");

  // The result may look like this:
  //
  // .. image:: ./move_group_interface_tutorial_avoid_path.gif
  //    :alt: animation showing the arm moving avoiding the new obstacle
  //
  // Attaching objects to the robot
  // ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  //
  // You can attach an object to the robot, so that it moves with the robot geometry.
  // This simulates picking up the object for the purpose of manipulating it.
  // The motion planning should avoid collisions between objects as well.
  moveit_msgs::msg::CollisionObject object_to_attach;
  object_to_attach.id = "cylinder1";

  shape_msgs::msg::SolidPrimitive cylinder_primitive;
  cylinder_primitive.type = primitive.CYLINDER;
  cylinder_primitive.dimensions.resize(2);
  cylinder_primitive.dimensions[primitive.CYLINDER_HEIGHT] = 0.20;
  cylinder_primitive.dimensions[primitive.CYLINDER_RADIUS] = 0.04;

  // We define the frame/pose for this cylinder so that it appears in the gripper.
  object_to_attach.header.frame_id = move_group.getEndEffectorLink();
  geometry_msgs::msg::Pose grab_pose;
  grab_pose.orientation.w = 1.0;
  grab_pose.position.z = 0.2;

  // First, we add the object to the world (without using a vector).
  object_to_attach.primitives.push_back(cylinder_primitive);
  object_to_attach.primitive_poses.push_back(grab_pose);
  object_to_attach.operation = object_to_attach.ADD;
  planning_scene_interface.applyCollisionObject(object_to_attach);

  // Then, we "attach" the object to the robot. It uses the frame_id to determine which robot link it is attached to.
  // We also need to tell MoveIt that the object is allowed to be in collision with the finger links of the gripper.
  // You could also use applyAttachedCollisionObject to attach an object to the robot directly.
  RCLCPP_INFO(LOGGER, "Attach the object to the robot");
  std::vector<std::string> touch_links;
  touch_links.push_back("panda_rightfinger");
  touch_links.push_back("panda_leftfinger");
  move_group.attachObject(object_to_attach.id, "panda_hand", touch_links);

  visual_tools.publishText(text_pose, "Object_attached_to_robot", rvt::WHITE, rvt::XLARGE);
  visual_tools.trigger();

  /* Wait for MoveGroup to receive and process the attached collision object message */
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window once the new object is attached to the robot");

  // Replan, but now with the object in hand.
  move_group.setStartStateToCurrentState();
  success = (move_group.plan(my_plan) == moveit::core::MoveItErrorCode::SUCCESS);
  RCLCPP_INFO(LOGGER, "Visualizing plan 7 (move around cuboid with cylinder) %s", success ? "" : "FAILED");
  visual_tools.publishTrajectoryLine(my_plan.trajectory, joint_model_group);
  visual_tools.trigger();
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window once the plan is complete");

  // The result may look something like this:
  //
  // .. image:: ./move_group_interface_tutorial_attached_object.gif
  //    :alt: animation showing the arm moving differently once the object is attached
  //
  // Detaching and Removing Objects
  // ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  //
  // Now, let's detach the cylinder from the robot's gripper.
  RCLCPP_INFO(LOGGER, "Detach the object from the robot");
  move_group.detachObject(object_to_attach.id);

  // Show text in RViz of status
  visual_tools.deleteAllMarkers();
  visual_tools.publishText(text_pose, "Object_detached_from_robot", rvt::WHITE, rvt::XLARGE);
  visual_tools.trigger();

  /* Wait for MoveGroup to receive and process the attached collision object message */
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window once the new object is detached from the robot");

  // Now, let's remove the objects from the world.
  RCLCPP_INFO(LOGGER, "Remove the objects from the world");
  std::vector<std::string> object_ids;
  object_ids.push_back(collision_object.id);
  object_ids.push_back(object_to_attach.id);
  planning_scene_interface.removeCollisionObjects(object_ids);

  // Show text in RViz of status
  visual_tools.publishText(text_pose, "Objects_removed", rvt::WHITE, rvt::XLARGE);
  visual_tools.trigger();

  /* Wait for MoveGroup to receive and process the attached collision object message */
  visual_tools.prompt("Press 'next' in the RvizVisualToolsGui window to once the collision object disappears");

  // END_TUTORIAL
  visual_tools.deleteAllMarkers();
  visual_tools.trigger();

  rclcpp::shutdown();
  return 0;
}

觉得对您有帮助的，可以点个赞👍支持一下，谢谢各位！

因为淋过雨，所以想为别人撑把伞；因为踩过太多坑，所以想让喜欢机器人的同学们减少试错成本！

押波张飞

关注

21
点赞
踩
24

收藏

觉得还不错? 一键收藏
打赏
0
评论
【附带源码】机械臂MoveIt2极简教程（二）、move_group交互

在MoveIt中，最简单的交互接口就是通过调用MoveGroupInterface类来实现的。它为大多数机械臂操作提供了最简单的使用功能，比如在设置关节位置点或者目标点后，想要用机械臂抓取东西；在现实环境中增加障碍物，让机械臂能绕障运动等等。这些都可以通过MoveGroup类中实现的ros node节点来完成。首先咱们本节的实现效果，心中有了效果图，再回头看代码，更加能够理解其中含义，也不会枯燥。打开两个terminal第一个terminal输入另一个terminal输入之后出现rviz界面，如下图所
复制链接

扫一扫