You don't need the Jacobian! What I have done is to set up a transformation that maps a force applied to the tool, to: the resulting torques in the upper arms. This can be done by "simple" trigonometry. This is in fact more or less the same as setting up the Jacobian, the only difference is that it is much better, because it can be used to calculate the internal forces in the structure, and thereby securing that these forces never exceeds the limit.
Yeah, one Propeller would be a much better choice than three toy Arduinos. That would simplify things down so that your control software only has to speak to a single COM port. Programming would be very trivial as objects already exist for the servos and for the serial comms. Just a bit of glue. Propeller is MUCH more responsive and accurate, which could be a real big boon here. We're seriously considering building something like this as a project on our show. I like the materials used.
Hello. Sorry that I have not been very active answering the comments in here, but I have been quite busy lately. No, the arduinos are only used as drives, which means that these are feeded with setpoints (angles, 1 every 1/1000 sec). There is one arduino for every motor, and it is only performing the PID regulation. The PC is performing the "realtime" Path Planning, which is off course only "99% realtime", as path planning can only be performed for the path lying ahead (future).
@jcims I claim that I have made a complete dynamical model for the robot. This is almost true! All ROTARY parts are modelled, except for the rods making up the parallellograms. The weight of these are modelled, but not the moments of inertia of these. However, I assume that since the moments of inertia of these are very small, the effect is neglectable.
@jcims The moments of inertia of the "upper arms" are a part of the model, just as the the inertia of the gear box and the rotor of the motor are modelled. In fact, the motor rotor and gear wheels, are those moving parts of the robot, which uses BY FAR the most amount of available torque. This is because these parts are rotating at very high speeds. Consequently, low moments of inertia of these parts are vital. But all these dynamical effects are considered in the path planning :-)
Is it expensive to build? How large of a scale can it be constructed? Is 4x4 feet feasible? How accurate is the "hand"? Is the programming complex? Are there 3rd party programs with friendly UIs to simplify programming for this?
At the moment I am trying to further develop the robot to use AC Induction Motors (or brushless DC). I'm looking into the technologi called "Field Oriented Control" at the moment. Also I'm trying to find an appropriate "Power Stage", so I don't need to make the high voltage components myself. If anyone have experiences with these technologies, I would very much appreciate to get in contact with you.
what kind of programans did you use to program the robot? I want to build this type of robot for my faculty degree, and i want to learn the programs, can you guide me to some good materials? What program did you use in this video?
+Gieljan Vantyghem The additional 3 arms keep the platform at the end of the arms parallel to the floor (plane of the servos). If you remove the additional arms, the platform can rotate in three axis as well as translate in 3 axis (6 degrees of freedom). To control the rotation, you need another three inputs (see www.moog.com/products/motion-systems/motion-bases/ which we use at work) The extra links act the same as moving pairs of actuators together - without having to use another 3 servos.
You don't need the Jacobian! What I have done is to set up a transformation that maps a force applied to the tool, to: the resulting torques in the upper arms. This can be done by "simple" trigonometry. This is in fact more or less the same as setting up the Jacobian, the only difference is that it is much better, because it can be used to calculate the internal forces in the structure, and thereby securing that these forces never exceeds the limit.
Yeah, one Propeller would be a much better choice than three toy Arduinos. That would simplify things down so that your control software only has to speak to a single COM port. Programming would be very trivial as objects already exist for the servos and for the serial comms. Just a bit of glue. Propeller is MUCH more responsive and accurate, which could be a real big boon here.
We're seriously considering building something like this as a project on our show. I like the materials used.
Hello. Sorry that I have not been very active answering the comments in here, but I have been quite busy lately.
No, the arduinos are only used as drives, which means that these are feeded with setpoints (angles, 1 every 1/1000 sec). There is one arduino for every motor, and it is only performing the PID regulation.
The PC is performing the "realtime" Path Planning, which is off course only "99% realtime", as path planning can only be performed for the path lying ahead (future).
Have you considered implementing force sensing resistors? They are relatively inexpensive at ~$8 and can be used for consistent automatic calibration.
класс!! а я зараз лінейку роблю.н6аступний теж буду робота робити
@jcims I claim that I have made a complete dynamical model for the robot. This is almost true! All ROTARY parts are modelled, except for the rods making up the parallellograms. The weight of these are modelled, but not the moments of inertia of these. However, I assume that since the moments of inertia of these are very small, the effect is neglectable.
Hanzhen harmonic drive gear , strain wave reducer,
robot gear joint, over 30 years experience
@jcims The moments of inertia of the "upper arms" are a part of the model, just as the the inertia of the gear box and the rotor of the motor are modelled. In fact, the motor rotor and gear wheels, are those moving parts of the robot, which uses BY FAR the most amount of available torque. This is because these parts are rotating at very high speeds. Consequently, low moments of inertia of these parts are vital. But all these dynamical effects are considered in the path planning :-)
where is download the plane.Where you buy the motor?thanks
Will you make a large version to stack my firewood for me? :)
Great work! How much do inertial effects play into your path planning?
Is it expensive to build? How large of a scale can it be constructed? Is 4x4 feet feasible? How accurate is the "hand"? Is the programming complex? Are there 3rd party programs with friendly UIs to simplify programming for this?
At the moment I am trying to further develop the robot to use AC Induction Motors (or brushless DC). I'm looking into the technologi called "Field Oriented Control" at the moment. Also I'm trying to find an appropriate "Power Stage", so I don't need to make the high voltage components myself. If anyone have experiences with these technologies, I would very much appreciate to get in contact with you.
Hi awesome work! What is the length of the black rods? Thanks!
what kind of programans did you use to program the robot? I want to build this type of robot for my faculty degree, and i want to learn the programs, can you guide me to some good materials?
What program did you use in this video?
Are delta bots like these accurate enough for a pick'n'place? Or are they not ideal because of the tool head weight?
Hello !
Are the Arduinos computing the path, kinematics, etc. or they are just used as I/O blocks for the PC ?
Regards
Hello,
Can you tell me the kinematic calculations for speed? I want to build a robot like this, but I can't calculate the Jacobian. Can you help me?
Hi, I'm interested to know what kind of servants you have. Regards!
is the code for path planner available? Or math behind it?
Why does a delta printer always need 6 arms and why would it not work with 3?
+Gieljan Vantyghem The additional 3 arms keep the platform at the end of the arms parallel to the floor (plane of the servos). If you remove the additional arms, the platform can rotate in three axis as well as translate in 3 axis (6 degrees of freedom). To control the rotation, you need another three inputs (see www.moog.com/products/motion-systems/motion-bases/ which we use at work) The extra links act the same as moving pairs of actuators together - without having to use another 3 servos.
It can actually work with just 5, but then it makes it non-symmetrical with 3 identical parts/mechanism for no real value.
How to control the robot which software used
How would you do that please help me to do that
Und was kann der ausser schnarren?
i want g-code, please