The AS5147 magnetic absolute rotation sensor is perfect for measuring robot joint angles. The key here is absolute! A relative rotation sensor like an encoder is very cheap but when you need to always totally unquestionably need to know where your thing is at then you need absolute sensing. In this post I’m going to show you my test setup, wiring, and code so you can get started easy peasy.
Denavit–Hartenberg (DH) parameters are a convention used by robotics experts to mathmatically describe the size and range of a robot’s bones. Today we’re going to look at what they mean and how they’re used.
So you built a Spot Micro (or you want to) and – because you’re very clever – you read the instructions first. You probably noticed there’s a section where you have to build a PCB from scratch. For a lot of people, that’s beyond their skill level …for now. In the meantime, nothing should hold you back from building a sweet walking robot dog, amiright? Cool.
Here’s a board we used in the past for a stewart platform, which are a type of robot you can read about elsewhere on this site. It’s a good start for driving a robot dog. ELEC-0132 stewart platform board was designed for 11 PWM signals out and 6 analog values in. I’m betting that one of them can be used as an extra out signal to drive all 12 motors in the Spot Micro’s legs.
In Robot Overlord have a JScrollPane in Java Swing and inside it is a JPanel that contains other JPanels and so on and way WAY down at the bottom were the selectable elements like JTextField and JButton. When I hit the TAB key (or SHIFT+TAB) to move focus between components I want the newly focussed component to stay on screen – in other words, make the computer move the scroll bars just enough to bring the focus component into view in the Viewport.
// either call on child with zero offset
child.scrollRectToVisible(new Rectangle(child.getSize());
// or call on child's parent with child bounds
child.getParent().scrollRectToVisible(child.getBounds());
Which sounds great, but doesn’t work with nested items. JComponent.getSize() won’t return the absolute position of the component relative to the top-most JPanel in the Viewport. JComponent.getBounds() is slightly better, with an X and Y value relative to its parent.
The solution
Recursion to the rescue!
public void focusGained(FocusEvent e) {
// the thing that got focus
Component c = e.getComponent();
// the bounds of the element
Rectangle rec = c.getBounds();
// add up all the parent offsets until we get to the JScrollPane.
Container c0 = null;
Container c1 = c.getParent();
while( (c1!=null) && !(c1 instanceof JScrollPane) ) {
Rectangle r2 = c1.getBounds();
rec.x += r2.x;
rec.y += r2.y;
// don't forget the last component before the JScrollPane.
// we'll need it later.
c0 = c1;
c1 = c1.getParent();
}
((JComponent)c0).scrollRectToVisible(rec);
}
It would have been really nice if calling scrollRectToVisible() on the first component would work its way up the parent chain automatically, but until that beautiful day this is your workaround.
Today I’d like to show you how to use Gradient Descent to solve Inverse Kinematics for your robot arm with a practical example. I used GDIK to improve the movement planning of the SIXI robot arm and I was blown away by how much simpler it was to understand and implement. Read on to see how it works so good.
