Hello,

I'm trying to run these examples on my ubuntu 18.x VM. I managed to install pybullet and numpy.But when i run the test, i get this

Traceback (most recent call last): File "", line 1, in ImportError: No module named ss

Thanks for your help. When i try to run to run the kuka test, i also get some errors

pybullet build time: Mar 5 2019 09:25:16 Traceback (most recent call last): File "/usr/lib/python3.6/runpy.py", line 193, in _run_module_as_main "main", mod_spec) File "/usr/lib/python3.6/runpy.py", line 85, in _run_code exec(code, run_globals) File "/home/thermo/Documents/dev/ss-pybullet/examples/test_kuka_pick.py", line 5, in from pybullet_tools.kuka_primitives import BodyPose, BodyConf, Command, get_grasp_gen,
File "/home/thermo/Documents/dev/ss-pybullet/pybullet_tools/kuka_primitives.py", line 4, in from .pr2_utils import get_top_grasps File "/home/thermo/Documents/dev/ss-pybullet/pybullet_tools/pr2_utils.py", line 11, in from .utils import multiply, get_link_pose, joint_from_name, set_joint_position, joints_from_names,
File "/home/thermo/Documents/dev/ss-pybullet/pybullet_tools/utils.py", line 20, in from motion_planners.rrt_connect import birrt, direct_path ModuleNotFoundError: No module named 'motion_planners'

Thanks for your help, Michael

Hi,

I am trying to grasp small blocks in pybullet using the PR2 gripper. However, I notice that there are some artifacts that I encounter while actually grasping the objects. This leads to significant inconsistencies. I have attached a gif of one such artifact. Can someone confirm the source of this issue and possible solutions?

In the example below, the gripper is closed by position controlling to a pre-determined value. There are also cases, where the gripper does not properly grasp the object causing it to fall.

The PR2 model was taken from this repository.

I build pybullet and pybullet_plan on my Windows system. but when I run the 'test_kuka_pick.py' ,I got an error in 'utils.py' ,'No module named 'motion_planners'. I want to know how to deal with this error.

Hi Caelan :) I hope you're well!

I have a question about the arguments to IKFast inverse kinematics. I'm looking here:

https://github.com/caelan/pybullet-planning/blob/8953d2dbfd789d5681053eb4993b5ddc0d910a6b/pybullet_tools/ikfast/ikfast.py#L163

The argument that I don't understand is free_positions. I'm guessing that this has something to do with wanting multiple possible solutions from the solver. But what does the IKFast solver expect for that argument? Sorry that I can't find documentation on the IKFast side for this.

Thanks for your help!

Hello, Thanks for such a great library. I have a problem where I do not know the current joint config and I am required to get multiple Ik solutions of Franka panda given the current end-effector position and orientation. I looked up the ikfast.py file and there I found pybullet_inverse_Kinematics function . if I set max-attempts arg greater than 1 would I will be able to get multiple Ik solutions? If not can you help how can I get multiple Ik solutions then for such a problem? Thanks

Hi,

I'd like to build an ikFast module for a Fetch robot and use it in pybullet env.

Based on what you have for the franka_panda robot:

1. I created the ikfast solver for my fetch's arm via "ikmodel.autogenerate()" and get the files "ikfast.h, ikfast_fetch_arm.cpp, ikfast_fetch_arm.so";
2. I copied the python bindings stuff (line 12835 - 12995) in "ikfast_panda_arm.cpp" into my "ikfast_fetch_arm.cpp";
3. I also copied all relevant files into my workspace;
4. After running "compile_ikfast", it shows "ikfast module ikfast_fetch_arm imported successful"

I noticed you set a free_joint for panda in ik.py, I was wondering what this free_joint means?

If I leave it as an empty list, I got an error saying "get_ik function takes exactly 3 arguments (2 given)" for line 26 shown below:

If I set the free_joint as "wrist_roll_joint", I always get none solution. Is this because I didn't set correct free_joints? if so do you have any idea what the correct joint to be set? Thank you very much!

I also attached the output of dump_body(robot) below:

If a trajectory with 2 consecutive duplicate states exists, then the current implementation of decompose_into_paths will consider this a new joint_group with 0 joints since none of the joints is manipulated. The proposed fix simply removes the duplicate consecutive state.

This PR contains the following modifications:

1. I get everything that is related to construction problem instances out of this repo.
2. I add the variable joints=conf2.joints in some planning fns (get_free_motion_gen, get_holding_motion_gen) in kuka_primitive. They assume all the movable joints by default, which is not the case for eth_rfl dual arms robot.

So after you merge them (by git or by hand), I shouldn't need to modify this submodule anymore.

Making PR2 TAMP example work in python 3.10 as mentioned in https://github.com/caelan/pddlstream/pull/35#issuecomment-1154658107 Require commit f067901bfabd3cd1a21ae42ef2088733eb1da476 in PR https://github.com/caelan/pddlstream/pull/35 to work