The MFR-LQ project I manage has completely overcome the problem of getting angles and is now the perfect version.

18 views
Skip to first unread message

takai hironori

unread,
Oct 4, 2020, 1:57:11 PM10/4/20
to ode-users
The project of MFR-LQ (Model Free Recursive LQ Controller) [sourceforge.net, OSDN], a software that uses control theory, has been completed.

The correct relative angle of the joint can now be obtained by integrating the relative angle of the objects connected by the joint with the value of dJointGetHingeAngleRate(); Now you have the code to get the relative angles in the range -infinite <--> +infinite.

The absolute angle of the correct object can now be obtained by integrating the value of the function of dBodyGetAngularVel(); with the absolute angle of the object, converting it to the difference of the angles, and adding it up. Now you have the code to get the absolute angle in the range -infinite <--> +infinite.

Thank you to everyone who answered the questions on the mailing list.

We are now able to release the official, perfect version of 3.4, which is sunny and bug-free.

From now on, I will start preparing to release the source code of 32bit and 64bit versions of Windows while repeating maintenance to perfect the source code of MFR-LQ.

By the way, we will also proceed with the work of porting to Linux in parallel. I would also like to create 32bit and 64bit versions of this as well.

If anyone is interested in control theory and robot simulation using physics engine and GSL, please email us.

We are looking for people who will cooperate and participate in the development.

I will do my best to port it to Linux.

By the way, the simulation implemented by MFR-LQ is
01. Simulation of a rotary inverted pendulum.
02. Simulation of rotary double inverted pendulum.
03. Single Wheel vehicle simulation and inertia rotor simulation.
04. Simulation of bogie type inverted pendulum (during implementation). There is a bug that digs into the floor.
05. It should also be useful for simulating robots with more complex models. Please improve if you are interested.

by hironori takai

e-mail: hironori....@gmail.com
e-mail: hironori...@cock.li



Reply all
Reply to author
Forward
0 new messages