Add inverse dynamics / mass matrix code from DeepMimic, thanks to Xue Bin (Jason) Peng
Add example how to use stable PD control for humanoid with spherical joints (see humanoidMotionCapture.py)
Fix related to TinyRenderer object transforms not updating when using collision filtering
See createMesh.py for an example.
The data has to fit in shared memory, hence the limit on Mac is lower than Windows and Linux:
#ifdef __APPLE__
#define B3_MAX_NUM_VERTICES 8192
#define B3_MAX_NUM_INDICES 32768
#else
#define B3_MAX_NUM_VERTICES 131072
#define B3_MAX_NUM_INDICES 524288
#endif
Made necessary changes to ```humanoid_deep_mimic_env.py``` file, setting ```initial_frame``` to new position after simulation steps and setting ```SimTime``` using this new initial frame time
Using calculateInverseDynamics with zero target acceleration allows to compute the non-linear dynamics forces (coriolis/gyroscopic) and/or gravity force.
implement deepmimic reward function (only joint angles/velocities for now)
fix an out-of-date comment, related to contactPoint.m_lateralFrictionInitialized in btSequentialImpulseConstraintSolver.cpp
Add preparation for DeepMimic humanoid environment, replicating parts of https://github.com/xbpeng/DeepMimic
Loading humanoid.urdf and applying motion action: examples/pybullet/gym/pybullet_envs/mimic/humanoid.py
Loading MotionCapture data: examples/pybullet/gym/pybullet_envs/mimic/motion_capture_data.py
Little test: examples/pybullet/gym/pybullet_envs/mimic/humanoid_test.py
Expose this btMultiBodySphericalJointMotor through PyBullet.setJointMotorControlMultiDof
Expose PyBullet.getQuaternionSlerp
Improve PyBullet.setJointMotorControlMultiDof
Improve humanoidMotionCapture.py with slerp and using setJointMotorControlMultiDof
Expose btMultiBody::spatialTransform
Fix btMultiBody::setupPlanar from DeepMimic codebase
Add support for multidof joints in btMultiBody::compTreeLinkVelocities, thanks to DeepMimic codebase @xbpeng
(mouse dragging adds energy in the system that can cause explosions/instability):
use solverResidualThreshold and increase number of sub steps, also clamp the max force.
If you use the Laikago URDF, please add a citation to PyBullet:
@MISC{coumans2018,
author = {Erwin Coumans and Yunfei Bai},
title = {PyBullet, a Python module for physics simulation for games, robotics and machine learning},
howpublished = {\url{http://pybullet.org}},
year = {2016--2018}
}
See createObstacleCourse.py for an example use. At the moment a limit of 1024 vertices and 1024 indices.
Will be lifted once we implement the streaming version (soon).
b3InitUpdateVisualShape doesn't take textureUniqueId by default. new API b3UpdateVisualShapeTexture to change texture (-1 will clear texture)
PyBullet/BulletRobotics: allow to reset the textureUniqueId to -1, to clear a texture
PyBullet/BulletRobotics: save all texture handles
enable follow cam in other Gym locomotion environments
add testing assets for multi-material obj files -> sdf conversion.
Also use ER_NO_SEGMENTATION_MASK flag for TinyRenderer/EGL plugin renderer
Support SDF loading through fileIO plugin.
Replace strcspn by C code (not crossplatform)
Add flag for loadURDF to use color from MTL file (instead from URDF link material)
pybullet.URDF_USE_MATERIAL_COLORS_FROM_MTL and pybullet.URDF_USE_MATERIAL_TRANSPARANCY_FROM_MTL
If parentObjectUniqueId provided, convert local from/to into world space coordinates
AddUserDebugLins: don't block when replacing an item
Fix examples/pybullet/examples/inverse_kinematics.py
PyBullet Allow OpenGL/EGL hardware to render segmentation mask. Use pybullet.ER_SEGMENTATION_MASK_OBJECT_AND_LINKINDEX or pybullet.ER_SEGMENTATION_MASK
PyBullet.removeBody fix indexing bug (use foundIndex, not i)
PyBullet bump up version to 2.2.3
Apply clang-format-all.sh using the _clang-format file through all the cpp/.h files.
make sure not to apply it to certain serialization structures, since some parser expects the * as part of the name, instead of type.
This commit contains no other changes aside from adding and applying clang-format-all.sh
allow to perform a getClosestPoints query with a collisionShape and world transform (position, orientation) that isn't part of the world.
(use createCollisionShape to create it)
add optional removeCollisionShape, for collision shapes only used in a query (and not used to create a body)
allow to perform a getClosestPoints query with a collisionShape and world transform (position, orientation) that isn't part of the world.
(use createCollisionShape to create it)
while (1):
stepSimulation()
pybullet.configureDebugVisualizer(pybullet.COV_ENABLE_SINGLE_STEP_RENDERING, 1)
disable single step using
pybullet.configureDebugVisualizer(pybullet.COV_ENABLE_SINGLE_STEP_RENDERING, 0)
Implement collisionFilterPlugin, use setCollisionFilterPair to enable or disable collision detection between specific pairs of objects.
Also, expose setCollisionFilterGroupMask as PyBullet API and in urdf using the tag <collision group="1" mask="2"/>.
See examples/pybullet/examples/collisionFilter.py for an example.
PyBullet default: Lower the warmstarting factor, for maximal coordinates rigid bodies for more stable simulation.
Add btCollisionWorld::refreshBroadphaseProxy to easier recreate the broadphase proxy without adding/removing objects to the world.
move setup.py back to eglRenderer extension, use pkgutil.get_loader('eglRenderer').get_filename()
disable dlmopen by default, unless B3_USE_DLMOPEN is defined.
just use pybullet.loadPlugin("eglRendererPlugin") before loading/creating any objects.
use end/startRendering (swap buffers) before the rendering in the eglPlugin
add grpcPlugin, it can work in GUI, SHARED_MEMORY_SERVER, DIRECT and other modes.
example script to start server from pybullet:
import pybullet as p
p.connect(p.GUI)
#if statically linked plugin
id = p.loadPlugin("grpcPlugin")
#dynamics loading the plugin
#id = p.loadPlugin("E:/develop/bullet3/bin/pybullet_grpcPlugin_vs2010_x64_debug.dll", postFix="_grpcPlugin")
#start the GRPC server at hostname, port
if (id>=0):
p.executePluginCommand(id, "localhost:1234")
Only in DIRECT mode, since there is no 'ping' you need to call to handle RCPs:
numRPC = 10
while (1):
p.executePluginCommand(id, intArgs=[numRPC])
This removes the need to specify the body id/link index when retrieving a user data entry.
Additionally, user data can now optionally be set to visual shapes as well.
The following public pybullet APIs have changed (backwards incompatible)
addUserData and getUserDataId
Makes linkIndex parameter optional (default value is -1)
Adds optional visualShapeIndex parameter (default value is -1)
getUserData and removeUserData
Removes required parameters bodyUniqueId and linkIndex
getNumUserData
Removes required bodyUniqueId parameter
getUserDataInfo
Removes required linkIndex parameter
Changes returned tuple from (userDataId, key) to (userDataId, key, bodyUniqueId, linkIndex, visualShapeIndex)
use b3RaycastBatchAddRays API to enable MAX_RAY_INTERSECTION_BATCH_SIZE_STREAMING num rays.
Old API (b3RaycastBatchAddRay) sticks to 256 rays, MAX_RAY_INTERSECTION_BATCH_SIZE.
reduce 'm_cooldownTime' from 1000 microseconds to 100 microseconds (overhead in raycast is too large)
If needed, we can expose this cooldown time.
Replace malloc by btAlignedObjectArray (going through Bullet's memory allocator)
Method def isAlive(self), which defaults to return self._alive < 0, and each environment can override this method (Half Cheetah would implement return False)
(In response to bea468fb93)
As suggested in https://github.com/bulletphysics/bullet3/pull/1759. The default isDone lets done = alive<0, and a special case is made for halfcheetah, forcing done=False.
I had to pass the 'alive' condition as an additive parameter of WalkerBaseBulletEnv.
To enable the feature, enable the BULLET2_MULTITHREADING option.
Increases the number of rays that can go in a batch request by storing
them in the shared memory stream instead of the shared memory command.
Adds the API b3RaycastBatchSetNumThreads to specify the number of
threads to use for the raycast batch, also adds the argument numThreads
to the pybullet function rayTestBatch.
Rays are distributed among the threads in a greedy fashion there's a shared
queue of work, once a thread finishes its task, it picks the next
available ray from the task. This works better than pre-distributing the
rays among threads, since there's a large variance in computation time per ray.
Some controversial changes:
- Added a pointer to PhysicsClient to the SharedMemoryCommand struct, this
was necessary to keep the C-API the same for b3RaycastBatchAddRay, while
adding the ray to the shared memory stream instead of the command
struct. I think this may be useful to simplify other APIs as well, that
take both a client handle and a command handle.
- Moved #define SHARED_MEMORY_MAX_STREAM_CHUNK_SIZE to
SharedMemoryPublic. This was necessary for the definition of
MAX_RAY_INTERSECTION_BATCH_SIZE.
Extract faces directly from btConvexHullComputer (in initializePolyhedralFeatures), instead of reconstructing them, thanks to Josh Klint in #1654
PyBullet: use initializePolyhedralFeatures for convex hulls and boxes (to allow SAT)
PyBullet: expose setPhysicsEngineParameter(enableSAT=0 or 1) to enable Separating Axis Test based collision detection for convex vs convex/box and convex versus concave triangles (in a triangle mesh).
return Py_None if no user data found, to pass the test.
enable m_deterministicOverlappingPairs by default.
if m_deterministicOverlappingPairs, sort the collision pairs.
Split examples/SharedMemory/b3RobotSimulatorClientAPI_NoGUI.* and move to examples/SharedMemory/b3RobotSimulatorClientAPI_NoGUI.cpp and examples/SharedMemory/b3RobotSimulatorClientAPI_NoDirect.cpp
Adds unit test for the UserData functons.
Changes the char pointer in btHashString to std::string. There were
problems where the object owning the string memory would deallocate the
string, making the btHashString object invalid.
allow to provide current joint positions in IK, overriding the body joint positions, also IK target will be in local coordinates.
expose b3ComputeDofCount in C-API
See https://github.com/erwincoumans/pybullet_robots ANYmal.py for an example.
PyBullet: Expose p.setPhysicsEngineParameter(solverResidualThreshold=1e-2) (b3PhysicsParamSetSolverResidualThreshold), increases solver performance a lot
PyBullet: Expose p.setPhysicsEngineParameter(contactSlop) Set it to zero, to avoid issues with restitution.
PyBullet: Expose isNumpyEnabled, return True is PyBullet was compiled with NUMPY support for 'getCameraImage'.
PyBullet: Expose p.ChangeDynamics(objectUid, linkIndex, contactProcessingThreshold), to avoid issues of speculative/predictive contacts with restitution.
See also http://twvideo01.ubm-us.net/o1/vault/gdc2012/slides/Programming%20Track/Vincent_ROBERT_Track_ADifferentApproach.pdf
use p.JOINT_FEEDBACK_IN_JOINT_FRAME if you want the joint feedback expressed in joint frame (instead of link inertial frame)
use p.JOINT_FEEDBACK_IN_WORLD_SPACE if you want the joint feedback in world space coordinates, instead of local link/joint coordinates.
Example: p.setPhysicsEngineParameter(jointFeedbackMode=p.JOINT_FEEDBACK_IN_WORLD_SPACE+p.JOINT_FEEDBACK_IN_JOINT_FRAME)
See also data\toys\concave_box.urdf and data\toys\concave_box.cdf
data\toys\concave_box.cdf was generated from concave_box.obj using
//GenerateSDF.exe -r "32 32 32" -d "-1.6 -1.6 -.6 1.6 1.6 .6" concave_box.obj
//SDF is based on code from DiscreGrid, https://github.com/InteractiveComputerGraphics/Discregrid
