Relationship between navigation packages
Four configuration files are required before the move_base node runs. These files define a series of related parameters, including the cost of overcoming obstacles, the radius of the robot, how long the future needs to be considered when planning the path, and how many blocks we want the robot to use. Speed moves and more. The four configuration files are:
base_local_planner_params.yaml
costmap_common_params.yaml
global_costmap_params.yaml
local_costmap_params.yaml
The navigation package uses a costmap to store information about obstacles. In order to do this correctly, we need to specify to the costmaps that the sensor topics should be listened to and how often they should be updated.
Reference article:
Run navigator:
1. Start the chassis control package:
base_conctronl2. Start the keyboard control node:
rosrun teleop_twist_keyboard teleop_twist_keyboard.py3. The robot side starts the radar scanning and mapping:
roslaunch robot_navigation robot_slam_laser.launch4. Start the Rviz view on the PC side
roslaunch robot_navigation slam_rviz.launch5. Enter the map package to save the package
roscd robot_navigation/maps #进包rosrun map_server map_saver -f map #保存地图，命名为map6. Start Radar Navigation:
roslaunch robot_navigation robot_navigation.launch7. Rviz navigation view:
roslaunch robot_navigation navigation_rviz.launch一、move_base.launch
<launch>
<!-- Arguments -->
<arg name="cmd_vel_topic" default="cmd_vel" /> # 话题控制
<arg name="odom_topic" default="odom" /> # 里程计
<arg name="planner" default="dwa" doc="opt: dwa, teb"/> # 导航
<arg name="simulation" default= "false"/> # 仿真
<arg name="use_dijkstra" default= "true"/>
<arg name="base_type" default= "$(env BASE_TYPE)"/>
<arg name="move_forward_only" default="false"/>
<!-- move_base use DWA planner-->
<group if="$(eval planner == 'dwa')">
<node pkg="move_base" type="move_base" respawn="false" name="move_base" output="screen">
<!-- use global_planner replace default navfn as global planner ,global_planner support A* and dijkstra algorithm-->
<param name="base_global_planner" value="global_planner/GlobalPlanner"/>
<param name="base_local_planner" value="dwa_local_planner/DWAPlannerROS" />
<!-- <rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/global_planner_params.yaml" command="load" /> -->
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/costmap_common_params.yaml" command="load" ns="global_costmap" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/costmap_common_params.yaml" command="load" ns="local_costmap" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/local_costmap_params.yaml" command="load" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/global_costmap_params.yaml" command="load" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/move_base_params.yaml" command="load" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/dwa_local_planner_params.yaml" command="load" />
<remap from="cmd_vel" to="$(arg cmd_vel_topic)"/>
<remap from="odom" to="$(arg odom_topic)"/>
<param name="DWAPlannerROS/min_vel_x" value="0.0" if="$(arg move_forward_only)" />
<!--default is True,use dijkstra algorithm;set to False,usd A* algorithm-->
<param name="GlobalPlanner/use_dijkstra " value="$(arg use_dijkstra)" />
</node>
</group>
<!-- move_base use TEB planner-->
<group if="$(eval planner == 'teb')">
<node pkg="move_base" type="move_base" respawn="false" name="move_base" output="screen">
<!-- use global_planner replace default navfn as global planner ,global_planner support A* and dijkstra algorithm-->
<param name="base_global_planner" value="global_planner/GlobalPlanner"/>
<param name="base_local_planner" value="teb_local_planner/TebLocalPlannerROS" />
<!-- <rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/global_planner_params.yaml" command="load" /> -->
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/costmap_common_params.yaml" command="load" ns="global_costmap" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/costmap_common_params.yaml" command="load" ns="local_costmap" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/local_costmap_params.yaml" command="load" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/global_costmap_params.yaml" command="load" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/move_base_params.yaml" command="load" />
<rosparam file="$(find robot_navigation)/param/$(env BASE_TYPE)/teb_local_planner_params.yaml" command="load" />
<remap from="cmd_vel" to="$(arg cmd_vel_topic)"/>
<remap from="odom" to="$(arg odom_topic)"/>
<param name="TebLocalPlannerROS/max_vel_x_backwards" value="0.0" if="$(arg move_forward_only)" />
<!--default is True,use dijkstra algorithm;set to False,usd A* algorithm-->
<param name="GlobalPlanner/use_dijkstra " value="$(arg use_dijkstra)" />
<!--stage simulator takes the angle instead of the rotvel as input (twist message)-->
</node>
</group>
<!-- move_base -->
</launch>1. teb_local_planner_params.yaml
This file is the parameters of TEB (the path planning algorithm used by NanoCar by default), and the parameters are detailed as follows:
TebLocalPlannerROS: # TEB局部路径规划
odom_topic: odom #使用里程计坐标
# Trajectory
teb_autosize: True #优化期间允许改变轨迹时域长度
dt_ref: 0.3 #局部路径规划解析度(0.01~1.0) 默认为0.3
dt_hysteresis: 0.1 #允许浮动范围
max_samples: 500
global_plan_overwrite_orientation: True #覆盖全局路径中局部路点朝向
allow_init_with_backwards_motion: True #允许开始后退回原来的执行的轨迹
max_global_plan_lookahead_dist: 3.0 #全局优化子集最大长度
global_plan_viapoint_sep: -1 #从全局计划中提取的每两个连续通过的点的最小间隔
global_plan_prune_distance: 1
exact_arc_length: False # ture则使用精确弧长
feasibility_check_no_poses: 2 # 检测位姿可达到的时间间隔
publish_feedback: False # 发布包含完整轨迹和障碍物
# Robotvb
max_vel_x: 0.8 # 最大前进速度 x
max_vel_x_backwards: 0.2 #最大后退速度 x
max_vel_y: 0.0 # y方向最大速度 阿克曼形是没有的
max_vel_theta: 0.53 # the angular velocity is also bounded by min_turning_radius in case of a carlike robot (r = v / omega) 最大转向角速度
acc_lim_x: 0.5 # 最大加速度
acc_lim_theta: 0.5 #最大角加速度
# ********************** Carlike robot parameters ********************
min_turning_radius: 0.375 # Min turning radius of the carlike robot (compute value using a model or adjust with rqt_reconfigure manually)——最小转弯半径
wheelbase: 0.145 # Wheelbase of our robot——驱动轴与转向轴之间的距离
cmd_angle_instead_rotvel: False # stage simulator takes the angle instead of the rotvel as input (twist message) —— 将收到的消息转化为操作
# ********************************************************************
footprint_model: # types: "point", "circular", "two_circles", "line", "polygon"
type: "line" # 设置机器人模型——可以为上面几种
line_start: [0.05, 0.0] # for type "line" #线的起点
line_end: [0.10, 0.0] # for type "line" #线的终点
# GoalTolerance
xy_goal_tolerance: 0.1 #xy目标偏移度
yaw_goal_tolerance: 0.2 #目标角度偏移容忍度
free_goal_vel: False #允许机器人以最大速度前往目的地，我们可以直接打开，然后设置一下我们上面的速度
complete_global_plan: True #完成目标点
# Obstacles
min_obstacle_dist: 0.12 # 与障碍物最小距离——与线
inflation_dist: 0.6 # 障碍物膨胀距离
include_costmap_obstacles: True # 局部地图中的障碍物是否考虑
costmap_obstacles_behind_robot_dist: 1.5
obstacle_poses_affected: 15
dynamic_obstacle_inflation_dist: 0.6 #动态障碍物膨胀范围
include_dynamic_obstacles: True #是否将为速度模型
costmap_converter_plugin: ""
costmap_converter_spin_thread: True
costmap_converter_rate: 5
# Optimization
no_inner_iterations: 5 # 被外循环调用后内循环执行次数
no_outer_iterations: 4 # 执行外循环优化次数
optimization_activate: True # 激活优化过程
optimization_verbose: False # 打印优化过程
penalty_epsilon: 0.1 # 对硬约束近似
obstacle_cost_exponent: 4
weight_max_vel_x: 2 # 最大速度权重
weight_max_vel_theta: 1 # 最大角速度权重
weight_acc_lim_x: 1 # 最大加速度权重
weight_acc_lim_theta: 1 # 最大角速度权重
weight_kinematics_nh: 1000
weight_kinematics_forward_drive: 1
weight_kinematics_turning_radius: 1
weight_optimaltime: 1 # must be > 0
weight_shortest_path: 0
weight_obstacle: 100 # 优化过程中和障碍物最小距离权重
weight_inflation: 0.2 # 膨胀区域权重
weight_dynamic_obstacle: 10 # 动态障碍物最小距离权重
weight_dynamic_obstacle_inflation: 0.2 # 动态障碍物膨胀区域权重
weight_viapoint: 1 #路径采样点距离权重
weight_adapt_factor: 2
# Homotopy Class Planner
enable_homotopy_class_planning: False
enable_multithreading: True # 允许多线程
max_number_classes: 4 # 允许线程数
selection_cost_hysteresis: 1.0 #
selection_prefer_initial_plan: 0.95
selection_obst_cost_scale: 1.0
selection_alternative_time_cost: False
roadmap_graph_no_samples: 15
roadmap_graph_area_width: 5
roadmap_graph_area_length_scale: 1.0
h_signature_prescaler: 0.5
h_signature_threshold: 0.1
obstacle_heading_threshold: 0.45
switching_blocking_period: 0.0
viapoints_all_candidates: True
delete_detours_backwards: True
max_ratio_detours_duration_best_duration: 3.0
visualize_hc_graph: False
visualize_with_time_as_z_axis_scale: False
# Recovery
shrink_horizon_backup: True # 规划检测系统异常，允许缩小时域规划范围
shrink_horizon_min_duration: 10
oscillation_recovery: False
oscillation_v_eps: 0.1
oscillation_omega_eps: 0.1
oscillation_recovery_min_duration: 10
oscillation_filter_duration: 10
2. move_base_params.yaml
This is the parameter of the move_base node itself, and it is also the basis for controlling other parameters
controller_frequency: 2.0 # 底盘发送数据
controller_patience: 15.0 # 控制器有效控制下发时间
planner_frequency: 0.2 # 全局规划执行频率
planner_patience: 5.0 # 留给规划期寻找路径的时间
conservative_reset_dist: 3.0 #
oscillation_timeout: 10.0 # 允许震荡时间
oscillation_distance: 0.2 # 移动多远才能发送不摆动
clearing_rotation_allowed: false # 是否使用恢复模块3. local_costmap_params.yaml
local_costmap: #局部代价地图
global_frame: map # 代价图的世界坐标系
robot_base_frame: base_footprint # 机器人局部坐标
update_frequency: 5.0 # 代价地图更新频率
publish_frequency: 5.0 # 代价地图发布频率
rolling_window: true # 是否使用动态窗口
width: 3 # 代价地图宽度
height: 3 # 代价地图宽度
resolution: 0.05 #精度
transform_tolerance: 0.5 #等待超时时间
plugins: # 固定使用，地图序列
- {name: static_layer, type: "costmap_2d::StaticLayer"}
- {name: obstacle_layer, type: "costmap_2d::ObstacleLayer"}
4. global_costmap_params.yaml
global_costmap: # 全局代价地图
global_frame: map # 地图名
robot_base_frame: base_footprint #机器人底部坐标系
update_frequency: 1.0 # 代价地图更新频率
publish_frequency: 0.5 # 发布频率
# static_map: true #设置为静态地图
transform_tolerance: 0.5 #等待超时时间
plugins:
- {name: static_layer, type: "costmap_2d::StaticLayer"}
- {name: obstacle_layer, type: "costmap_2d::VoxelLayer"}
- {name: inflation_layer, type: "costmap_2d::InflationLayer"}
5. costmap_common_params.yaml
footprint: [ [-0.035,-0.1], [0.18,-0.1], [0.18,0.1], [-0.035,0.1] ] #机器人占有面积
transform_tolerance: 0.2 # 等待坐标发布超时时间
obstacle_layer: # 障碍层
enabled: true
obstacle_range: 2.5 #放入代价地图的距离
raytrace_range: 3.0 #清除代价地图的距离
inflation_radius: 0.05 #膨胀半径
track_unknown_space: false
combination_method: 1
observation_sources: laser_scan_sensor #传感器参数源：雷达
laser_scan_sensor: {data_type: LaserScan, topic: scan, marking: true, clearing: true}
#设置源消息 消息类型 发布话题 障碍物消息 是否清除障碍物消息
inflation_layer: #膨胀层
enabled: true #使能
cost_scaling_factor: 10.0 # 代价比例 越大价值越小(default: 10)
inflation_radius: 0.1 # 膨胀半径
static_layer: #静态层
enabled: true
map_topic: "/map" #话题名Modifications on parameters:
In the current version, we are using the Jetson nano Ackerman series, and we are using the basic version, so the maximum speed when we measure the speed should be 1.5m, and we set it to 0.8m for testing.
Summary reference for performance issues of TEB planner
· Turn off multipath parallel planning (the effect is very significant)
· Use Costmap Converter (very significant)
· Reduce the number of iterations (no_inner/outer_iterations) (significant)
· Reduce max_lookahead_distance (general)
· Reduce the size of the local cost map (significant)
· Increase Large planning cycle and control cycle (affecting effect)
Use a single point footprint, with the minimum obstacle distance constraint (not too significant and affect the effect)
We mainly refer to the teb.yaml parameter table, which can directly affect our speed.
dt_ref 局部路径规划的解析度（默认0.3）
dt_hysteresis 允许改变的时域解析度的浮动范围， 一般为 dt_ref 的 10% 左右（默认0.1）dt_ref and dt_hysteresis Default distance of two adjacent attitudes (ie position, velocity, heading information, can be seen in rivz via TEB visualization) on the optimal path.
This distance is not fixed, and the planner automatically adjusts this distance according to the speed. The greater the speed, the greater the adjacent distance.
global_plan_overwrite_orientation 覆盖全局路径中局部路径点的朝向，有些全局规划者在开始和全局目标之间没有考虑局部子目标的方向，因此自动确定In our example, we are enabled, in order to find multiple routes to the destination, although this can reduce our collision, but it will reduce our speed, for the 2D planning of the vehicle, it can be set to False, which can be realized Better tracking of global paths.
max_global_plan_lookahead_dist
The maximum length (cumulative Euclidean distance) of the subset of global plans considered for optimization (if 0 or negative: disabled; length is also limited by the size of the local Costmap), default 3.0
This parameter determines the maximum length of the initial trajectory of the local planning. The actual debugging found that this parameter does not need to be too large, because the local trajectory is updated in each control cycle, and the actual executed command is only the speed value of the first point on the trajectory. Here, set It is enough to be 1.5 (default 3.0), too long may also cause the optimization result to fail to converge effectively. We made it smaller in our example.
Requirements: 1. It should increase with the increase of the maximum speed of the vehicle. 2. It should not exceed the reliable measurement range of sensors such as laser radar. to plan.
feasibility_check_no_poses 检测位姿可到达的时间间隔
It is used when judging whether the generated trajectory conflicts. At this time, it is set to 3, that is, the three points on the trajectory are checked one by one from the starting point of the trajectory. If there is no collision between the three points, it is considered that the trajectory is valid, because teb optimization is not Hard constraint, here is equivalent to a layer of guarantee after trajectory generation, this parameter is adjusted according to the speed of the robot and the complexity of the environment, otherwise it is very likely to stop and go in a narrow environment.
max_vel_x acc_lim_x x轴方向速度与加速度约束
max_vel_y acc_lim_y y轴方向速度与加速度约束
max_vel_x_backwards 最大倒车速度
Ackerman's y is 0, the maximum speed of x we said above, set it to 1, and we can set the acceleration to 0.3.
Note that the acceleration also constrains the deceleration process. If the motor performance is good and the speed is not fast, you can brake immediately. You can directly set acc_lim_x to 0, which means there is no constraint. If the motor cannot withstand step input or the response time is very long, the acceleration limit should be set.
Prohibiting reversing should weight forward travel very high in the penalty section, but we don't need to prohibit it.
The GoalTolerance parameter sets the tolerance for the robot to stop running
xy_goal_tolerance和yaw_goal_tolerance 目标跟踪误差Set according to the vehicle running accuracy. For example, the author uses a climbing RC car to build a navigation system. This kind of vehicle has a large steering clearance, so strict heading restrictions should not be set
free_goal_vel 自由目标速度In the racing part, we can directly set it to True in the parameter to let the car move forward at full speed
The Obstacles parameter is used to deal with obstacles in the environment, which is reflected in the trajectory optimization stage
costmap_converter_plugin 是否使用costmap_converter插件 若设置为空字符，则视为禁用转换
The role of the costmap_converter plugin is to pre-represent obstacles in the form of line segments or polygons, which can reduce the pressure of subsequent distance calculations to a certain extent. For details, see costmap_converter - ROS Wiki
include_costmap_obstacles 必须设置为True后才能规避实时探测到的、建图时不存在的障碍物。
This is what we have to open
min_obstacle_dist 最小障碍物距离
Increasing min_obstacle_dist can prevent the robot from getting too close to the wall, but for narrow passages, it should be set very small instead. The settings of these three parameters are very important and need to be carefully adjusted according to the robot's external dimensions, otherwise it is very easy for the robot to fail to pass through in a narrow space or the optimization does not converge.
inflation_dist 障碍物膨胀距离
The default is 0.0 for obstacle expansion distance. This value must be greater than min_obstacle_dist to be valid. The source code is in AddEdgesObstacles. This inflation only lowers the priority of passing through these areas, and this distance should not be used to force the vehicle to stay away from obstacles. The buffer around the obstacle, zero penalty cost, the buffer will cause the planner to slow down, we set it to 0.6
costmap_obstacles_behind_robot_dist double 限制在机器人后面规划时考虑到的占用的本地成本图障碍(指定距离，单位为米)
obstacle_poses_affected int 障碍物位置与轨迹上最接近的姿态相连，以减少计算量，但同时也考虑了许多相邻的障碍物
https://www.jianshu.com/p/a0f1d93f5179
The Optimization parameter is mainly to set the weight of each part in the optimization framework
We give reference link here
no_inner_iterations int 被外循环调用后内循环执行优化次数 1 5 100
no_outer_iterations int 执行的外循环的优化次数 1 4 100
optimization_activate bool 激活优化 False True True
optimization_verbose bool 打印优化过程详情 False False True
penalty_epsilon double 对于硬约束近似，在惩罚函数中添加安全范围 0.0 0.1 1.0
weight_max_vel_x double 最大x速度权重 0.0 2.0 1000.0
weight_max_vel_y double 最大y速度权重 0.0 2.0 1000.0
weight_max_vel_theta double 最大加速度权重 0.0 1.0 1000.0
weight_acc_lim_x double 最大x 加速度权重 0.0 1.0 1000.0
weight_acc_lim_y double 最大y 加速度权重 0.0 1.0 1000.0
weight_acc_lim_theta double 最大角速度权重 0.0 1.0 1000.0
weight_kinematics_nh double 满足非完整运动学的优化权值 0.0 1000.0 10000.0
weight_kinematics_forward_drive double 优化过程中，迫使机器人只选择前进方向，差速轮适用 0.0 1.0 1000.0
weight_kinematics_turning_radius double 优化过程中，车型机器人的最小转弯半径的权重 0.0 1.0 1000.0
weight_optimaltime double 优化过程中，基于轨迹的时间上的权重 0.0 1.0 1000.0
weight_obstacle double 优化过程中，和障碍物最小距离的权重 0.0 50.0 1000.0
no_inner_iterations: 5 图优化optimizer的迭代次数
no_outer_iterations: 4 每次外部循环迭代都会根据所需的时间分辨率dt_ref自动调整轨迹的大小，并调用内部优化器
penalty_epsilon 0.1 一次性改变所有的惩罚项，为惩罚函数增加一个小的安全余量，以实现硬约束近似. 例如为速度的约束提供一个缓冲的效果，在到达速度限制前会产生一定的惩罚，让其提前减速达到缓冲的效果
weight_obstacle 默认值：50.0. 优化权重以保持与障碍物的最小距离。可以增大至几百，让机器人提前转向，避免卡死
obstacle_cost_exponent 无默认值，目前设置为0.65. 源码在EdgeObstacle::computeError和EdgeInflatedObstacle::computeError，判断是否为1，同时min_obstacle_dist > 0才有效。用于更新g2o的误差函数
weight_kinematics_nh 默认值：1000. 用于满足非完整运动学的优化权重. 该参数必须很高，因为运动学方程构成了一个等式约束，即使值1000也不意味着由于与其他成本相比较小的“原始”成本值而导致的矩阵条件不佳
Below I will put some parameter descriptions on the official website, and then take a closer look when adjusting the parameters
1. two/teb_planner