Search found 73 matches

by steven
Tue Nov 26, 2019 12:52 am
Forum: General Bullet Physics Support and Feedback
Topic: Collision detection failed for Ellipsoid shapes
Replies: 7
Views: 576

Re: Collision detection failed for Ellipsoid shapes

temp = (*pos) + m_localScaling * (vec * m_localScaling).normlized() * (*rad) - vec * getMargin(); feel it should be like this: temp = (*pos) + m_localScaling * (vec / m_localScaling).normalized() * (*rad) - vec * getMargin(); i will check if the points on the ellipsoid can be expressed by this func...
by steven
Mon Nov 25, 2019 2:38 am
Forum: General Bullet Physics Support and Feedback
Topic: Collision detection failed for Ellipsoid shapes
Replies: 7
Views: 576

Re: Collision detection failed for Ellipsoid shapes

adding this shape->setMargin(0) can avoid this issue but i still can't understand what caused this issue. i have some questions when review this code below in line 76 of btMultiSphereShape.cpp. temp = (*pos) * m_localScaling + vec * m_localScaling * (*rad) - vec * getMargin(); 1. why the origin pos ...
by steven
Thu Nov 14, 2019 7:14 am
Forum: General Bullet Physics Support and Feedback
Topic: What parameters to tune to stabilize collisions
Replies: 8
Views: 710

Re: What parameters to tune to stabilize collisions

hi JJJK,
i still suspect that this is caused by the non-suitable actions. i find the every joint of the human description with URDF format doesn't have any limitation while in the MJCF format it does.
hope you can update your latest demo.
thanks.
by steven
Tue Nov 12, 2019 7:53 am
Forum: General Bullet Physics Support and Feedback
Topic: What parameters to tune to stabilize collisions
Replies: 8
Views: 710

Re: What parameters to tune to stabilize collisions

hi all,
i have another question not specific to this case, where does the motor constraint force come from? this force will increase the system energy? thanks.
by steven
Tue Nov 12, 2019 12:55 am
Forum: General Bullet Physics Support and Feedback
Topic: What parameters to tune to stabilize collisions
Replies: 8
Views: 710

Re: What parameters to tune to stabilize collisions

did you only change the robot description to the pybullet_data/mjcf/humanoid.xml then it can work now? did you have any other changes? could you also attach them also? thanks.
by steven
Mon Nov 11, 2019 9:37 am
Forum: General Bullet Physics Support and Feedback
Topic: What parameters to tune to stabilize collisions
Replies: 8
Views: 710

Re: What parameters to tune to stabilize collisions

looks not so reasonable. I will check it again later.
by steven
Mon Nov 11, 2019 9:00 am
Forum: General Bullet Physics Support and Feedback
Topic: What parameters to tune to stabilize collisions
Replies: 8
Views: 710

Re: What parameters to tune to stabilize collisions

i just take a test that change the gravity acceleration to setGravity(0, 0, -0.981) and set robot initial position to robot_init_pos=(0, 0, 200) and then after a while we can find that the base velocity sometimes become very large. i also debug this case and find the acceleration of the base and joi...
by steven
Thu Nov 07, 2019 11:40 am
Forum: General Bullet Physics Support and Feedback
Topic: What parameters to tune to stabilize collisions
Replies: 8
Views: 710

Re: What parameters to tune to stabilize collisions

thanks for your sharing. i think it should be caused by your selected actions which maybe make the robot jump very high.
by steven
Mon Oct 14, 2019 5:24 am
Forum: PyBullet Support and Feedback
Topic: Simulating an KUKA LBR iwaa
Replies: 2
Views: 448

Re: Simulating an KUKA LBR iwaa

Could you provide a complete simple test? thanks.
by steven
Wed Sep 25, 2019 3:36 am
Forum: General Bullet Physics Support and Feedback
Topic: Why timeStep have to be this unreasonably small?
Replies: 7
Views: 1586

Re: Why timeStep have to be this unreasonably small?

This is from the assumption of linear elasticity. Similar to Hooke's Law, the torque is promotional to how much it displaces from the equilibrium position (angle). i hava an intuition that the torque of joint near the base should be larger than the joint near the end given the same delta theta, rig...
by steven
Mon Sep 16, 2019 1:26 pm
Forum: General Bullet Physics Support and Feedback
Topic: Why timeStep have to be this unreasonably small?
Replies: 7
Views: 1586

Re: Why timeStep have to be this unreasonably small?

sorry, i still don't understand why this function be written like that. i have some questions as below. 1. where do these constants come from such as -1.5e-3, 1.5735*1e5? 2. why the joint torque only have relation with the joint delta theta(if not consider the damping)? also you can call the POSITIO...
by steven
Wed Sep 11, 2019 8:56 am
Forum: General Bullet Physics Support and Feedback
Topic: Why timeStep have to be this unreasonably small?
Replies: 7
Views: 1586

Re: Why timeStep have to be this unreasonably small?

i change your code to use the POSITION_CONTROL mode with (5e-3)/240 us timestep which is bigger than what you used and it can work well as below. if you have to use the TORQUE_CONTROL mode based on some other consideration, Could you explain your function of TorVolThe(Theta,angularVelocity,Voltage)?...
by steven
Thu Sep 05, 2019 2:27 am
Forum: General Bullet Physics Support and Feedback
Topic: Why timeStep have to be this unreasonably small?
Replies: 7
Views: 1586

Re: Why timeStep have to be this unreasonably small?

hi Could you tell me the bullet version you used? i am testing this case with the attached code but get a different result.
my bullet version is 2.5.5.
by steven
Tue Aug 27, 2019 12:58 am
Forum: General Bullet Physics Support and Feedback
Topic: unstable soft body similation
Replies: 3
Views: 912

Re: unstable soft body similation

Could you email these object files to me, shengjun89@163.com? thanks.