This is a ros workspace for running the reachability-aware grasping pipeline. It includes scripts for generating binary reachability space, converting to sdf representation and integrating it with the GraspIt! grasp planner.
- ROS Indigo
- GraspIt!
- Mongo DB! (Optional: Only needed if you want to run offline computation of reachability space)
- gitman
- scikit-fmm
git clone http://github.com/CRLab/reachability_aware_grasping_ws.git
gitman install
catkin_make
To see the reachability-aware grasping approach in simulation, using our precomputed reachability space for the Fetch robot, run the scripts as described [here] (https://github.com/CRLab/reachability_grasping_demo.git).
To do the offline computation your self, follow steps below.
- Define workspace dimension and resolution and upload grid points as tasks to database
cd src/reachability_space_generation/scripts/reachability_space_generation
python upload_reachability_tasks.py fetch22.yaml
- Compute Inverse Kinematic checks for grid points in the workspace
cd src/reachability_space_generation/launch
./launch_fetch_space_generation.sh
- Download IK Check result as csv
python download_mongo_db.py fetch_to_do fetch_finished fetch_dense_reachability.csv
This is done in the reachability_utils package. Copy the binary ik check result from previous step into a data folder in the reachability_utils package.
cd src/reachability_utils/scripts/reachability_utils
python process_reachability_data_from_csv.py fetch_dense_reachability.csv
The processed sdf data folder can be copied into a config package used for GraspIt! reachability_energy_plugin. Processed reachability space for the Fetch robot is available [here])(https://github.com/CRLab/fetch_reachability_config.git)