Manipulation of Deformable Linear Objects in Benchmark Task Spaces (IEEE CASE 2022)

Paper link

Video link

System overview:

System requirement:

ROS Kinectic

Darknet_ros

Opencv

PCL

Kinova ros_kortex package

Kinova ros_kortex_vision package

Trajop_ros

Hardware requirement:

Kinova gen3 arm mounted with RGBD camera

Introduction:

/darknet_ros folder includes all the configurates and weights for tube and cable detection

/nist folder includes all the launch file, source codes, and service files

/research folder includes all the configuration and launch files for integrating Trajopt with Gen3 arm

/trajopt_ros folder includes the trajopt (optimized motion planner) in ROS

Commands

--launch robot and camera

roslaunch nist nist.launch

--launch trajopt service node

roslaunch research gen3_trajopt_node.launch

--insert cable modeling (servoing) (insert to tube 1)

rosrun nist modeling_cable_nist_insert

--insert cable modeling (servoing) (insert to final holder)

rosrun nist modeling_cable_nist_insert_final_holder

--run trajopt - go the start position

rosrun nist nist_start __ns:=my_gen3

--run color detection node - kinova camera

rosrun nist color_detection_blue_kinova_node

--run modeling cable node for pushing cable

rosrun nist modeling_cable_gen3_nist_node

--run yolo detection to detect the board and tubes

roslaunch darknet_ros darknet_ros_NIST_board.launch

--run yolo detection to detect the cable

roslaunch darknet_ros darknet_ros_NIST_cable.launch

--run plane detection to detect the board plane

rosrun nist plane_segmentation

--run trajopt - run the task

rosrun nist nist_run __ns:=my_gen3

Copyright:

This work was developed at the RIVeR Lab, Northeastern University