Commit Graph

11 Commits

Author SHA1 Message Date
Erwin Coumans
d3d0d948cc data: add license files (mostly copied from the URDF files) 2018-08-09 15:50:39 -07:00
erwincoumans
0d8564db22 PyBullet: added preliminary DART and MuJoCo backend files, MuJoCo can loadMJCF, stepSimulation and getBasePositionAndOrientation, DART is empty.
PyBullet: add 2d biped example.,
2018-07-01 14:42:32 -07:00
erwincoumans
507d024f25 add example of a free moving sphere controlled by external torque
add some toy shape sorter cube (not finished yet)
2018-04-01 15:55:32 -07:00
Benelot
aed57d130e Update mjcf files!
----------------------------------
Git did not realize I changed those.
2017-08-17 00:04:21 +02:00
Erwin Coumans
fd40ba424b test for knee joint limits under position control 2017-05-21 11:00:15 -07:00
Erwin Coumans
79e2c10506 add second humanoid_running_3.py example with 3 humanoids, for testing. Created a humanoid_symmetric_no_ground.xml and ground.xml:
if the ground is duplicated, simulation is unstable (too many contacts in the same area?)
2017-05-12 17:17:27 -07:00
Erwin Coumans
385156cbc5 add simple humanoid_benchmark.py and mjcf file 2017-05-05 17:24:35 -07:00
Erwin Coumans
e32debdca4 improvements in MJCF importer: fix capsule shape inertia (for fromto definition)
add cylinder support
don't crash when no "quat" is provided
inertia fix in btMinkowskiSumShape (based on AABB)
2017-04-05 12:22:38 -07:00
Erwin Coumans
76dcf3a751 remove the contype=0 in inverted_pendulum.xml, so we can 'mouse-pick' the pendulum.
[mjcf importer] add rudimentary support for inertial frame computation, for 'fromto' capsules.
2017-01-24 21:10:21 -08:00
Erwin Coumans
310fabcd7e Further improvements in Bullet MuJoCo MJCF import:
[MJCF import] improved humanoid.xml, inverted_pendulum.xml import (automatically add dummy-links to connect multiple joints between bodies)
[MJCF import] fix bug in quaternion conversion (w,x,y,z) -> (x,y,z,w)
2017-01-15 11:22:24 -08:00
Erwin Coumans
fdd517e00f First step towards a MuJoCo MJCF importer for Bullet. It can load the humanoid.xml, ant.xml and some other OpenAI GYM asset files. Not all fields are converted, so it is work-in-progress. This is useful for Reinforcement Learning experiments, and would also help integration with DeepMind Lab. 2016-12-30 18:32:57 -08:00