YeeKal / grasp_ws

grasp workspace for the master thesis based on ros

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

The box grasping workspace for master thesis.

Content:

help command

'''
realsense
'''
## driver
roslaunch realsense2_camera rs_camera.launch filters:=pointcloud align_depth:=true

rosrun pcl_tutorial save_depth_with_mask /camera/color/image_raw /camera/depth/image_rect_raw imgs


# main
rosrun pcl_tutorial main reso/box_small/thin.pcd -o 0.7 -d 0.01 -t 1000 -n 100 -it 100

# laser line
roslaunch pcl_tutorial filter_laser.launch

roslaunch laserline laserline.launch 

roslaunch laser_line_extraction example.launch
rosrun pcl_tutorial counter_pose 2.0 0.65,0.75
# tf
rosrun tf static_transform_publisher  0.00662214  0.0339903 -0.0687947  0.919409 -0.392716 0.0021969 0.0213767 arm_left_link_7_t camera_rgb_optical_frame 100

## sda driver
roslaunch motoman_sda5f_support robot_interface_streaming_sda5f.launch 
roslaunch motoman_sda5f_moveit_config demo_real.launch 

technique

  1. laser detection
    • laser points differ
    • low-pass filter
    • line detection
  2. camera recognition
    • hsv segmentation
    • 6D pose estimation: ICP+ Super4PCS
  3. motion planning

Experiment

time

constraint:

  • moveit: p:0.03009/execute:5.8763 0.03543/5.9266123
  • yeebot: p:0.024/4.337 0.024/4.56
  • conclusion:
    • planning time com: 0.035-0.024
    • execution com: 4.337-5.926
back cartesian constraint counter block
planning 0.0183 0.0246 0.0219 0.0356 0.5130
execution 3.79 6.86 3.62 - -
all time 14.8884 planning 0.0648 execution 14.2750

241/h

hardware

sda

initial states:

[arm_left_joint_1_s, arm_left_joint_2_l, arm_left_joint_3_e, arm_left_joint_4_u, arm_left_joint_5_r,arm_left_joint_6_b, arm_left_joint_7_t]
position: [2.6937308311462402, -1.7499834299087524, -1.922503113746643, 0.04039989039301872, 0.8331186175346375, -1.1421606540679932, 1.8852884769439697]

initial_states(2019-12-05):

[arm_left_joint_1_s, arm_left_joint_2_l, arm_left_joint_3_e, arm_left_joint_4_u, arm_left_joint_5_r,

arm_left_joint_6_b, arm_left_joint_7_t] position: [-0.022554073482751846, 1.3352922201156616, 1.1472201347351074, 0.19595465064048767, 0.7072410583496094, -0.9488791227340698, 1.834251046180725]

counter pose: 0.74654 4.07403e-17 -0.66534 0.76544 -0.66534 4.57124e-17 -0.74654 -0.963605 0 1 6.12323e-17 0 0 0 0 1

laser

hokuyo: 40hz

ip

sda: 192.168.1.10 hokuyo: 192.168.1.11 pc-sda: 192.168.1.87

tf

The sensor link is published by static_transform_publisher in motoman_sda5f_moveit_config/launch/motoman_sda5f_moveit_sensor_manager.launch

sensor list: - asus camera: 0.00662214 0.0339903 -0.0687947 0.919409 -0.392716 0.0021969 0.0213767 /arm_left_link_7_t /camera_rgb_optical_frame 100 - hokuyo laser: 0.43 -0.15 0.68 0 0 0 1 /world /laser 100

model size

right gripper:

  • gripper legth: 210mm
  • gripper bottom to end-effector: 134.7mm
  • gripper finger width: 5mm

camera calibration

# asus 2
*** Added sample 152, p_x = 0.661, p_y = 0.708, p_size = 0.253, skew = 0.724
('D = ', [-0.010767126205035457, 0.02878932007658363, -0.007516792631713471, -0.003062193530694651, 0.0])
('K = ', [536.3783314976719, 0.0, 325.59688771858265, 0.0, 539.4028276436596, 240.028337450314, 0.0, 0.0, 1.0])
('R = ', [1.0, 0.0, 0.0, 0.0, 1.0, 0.0, 0.0, 0.0, 1.0])
('P = ', [541.3510131835938, 0.0, 323.52929686348944, 0.0, 0.0, 542.7696533203125, 236.24413879678468, 0.0, 0.0, 0.0, 1.0, 0.0])

dependecies

ros package inside

ros package extral

system package

pose estimation paras

items           time/ms   score/e-5
super4pcs+icp   479         3.264
                498         3.971
super4pcs       289         3.682
                241         3.98
                281         4.337
                688         3.45
super4pcs(before)212        3.9
                428         3.25
                374         4.62
super4pcs+icp   1460        2.73
                1380        2.82
                1040        2.75
ppf             400         k
                1020
                876

About

grasp workspace for the master thesis based on ros


Languages

Language:C++ 81.1%Language:C 10.0%Language:CMake 7.2%Language:Python 1.6%Language:MATLAB 0.1%Language:Shell 0.0%