Let's Make Robots!

Rhex

I definitely want to build a 6 legged Rhex robot. Below are some drafts I made today with Sketchup. 6 standard servos HS-325, a LiPo battery, two Sharp IR distance sensors and a tilt switch.

 


2011.2.14

I thought about how to synchronize the legs of my Rhex robot. I want to use servos, no kind of encoders and geared motors. But the servo would need to have two functions: Continuous rotating and moving to a desired position. So I came up with following idea:

The end stopf on the servo gear needs to be removed but the potentiometer spindle must be still connected to the motor. The potentiometer needs to be without end stop(s). At least some micro servos have this kind of potentiometer without end stop(s). If A=high, the servo rotates continuous, if A=low, the servo moves to a desired position. So far the theory.


2011.2.15

Built a prototype of the suggested servo, but it works not like designated. One problem is that the outer potentiometer pins are not directly connected to GND and VCC of the servo PCB. Back to the drawing board...

Dismantled a potentiometer from a cheap standard size servo. Just the marked end stop must be removed with a sharp knife and you have a continious rotating potentiometer. The values of one turn is from 0-5kΩ and on a very short distance the wiper contact is disconnected from the fixed contacts.


2011.2.16

Using now a relay to avoid all potential and voltage drop problems:

Ordered TAKAMISAWA A5W-K PCB relays. I have attached the data sheet. They are very tiny, just like a 10 pin DIL chip. And they are DIL socket compatible:

 


 

2011.2.17

Finally it works. Here is the current setup:

The modified now continious rotating potentiometer works fine. I'll make a short video at weekend. Instead of a relay I will use analog switches at the final version as Oddbot has suggested.

AttachmentSize
07-01-4005.pdf431.85 KB

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

can anyone tell me, where can i Buy these legs. I need them urgently to finish my project. Please Help.

Hi guys im new here and havn't even made my first robot yet :( anyhow I began this new interest when i was looking at the boston dynamics website and my jaw hit the flaw when i saw what these guys are doing and loved the rhex! Anyways i stumbled across a kids robot kit thats isn't realy a robot but its a rip off from the rhex! maybe you can even hack this thing and save some time! It has more like an s shaped paddel that it walks on maybe this is to solve thoes timing problems you guys ae talking about, anyhow heres the link....enjoy

oh and i would love to here about any developments you guys have come up with on this thing......love it, so simple but effective!

http://www.robotshop.com/roboxplorer-multi-terrain-robotic-toy-vehicle-2.html

I have always wanted to make one of these. I will have to see what you come up with :).

Hi Patrick,

I also will have to wait till DAGU manufactured my servo. This will make it mutch easier to build a Rhex. Are you interested in a collaborative project?

You really need to use FETs to switch your pot and resistors. I would suggest trying a CMOS IC such as a 74HC240. This is an octal tri-state buffer with each output rated for about 35mA. As this buffer has two groups of 4 with each group having it's own tri-state enable you can use one group to switch to ground and another to switch to Vcc.

I thought on FETs. But I still have a problem with the servo PCB. The voltage divider (potentiometer) is not connected directly to GND and VCC. One fixed connection of the potentiometer has nearly a potential of 0V to GND, the other fixed connection around 2.7V @ 5V VCC. A circuit diagram of the servo would be very helpful. I am also afraid of the voltage drop by transistor/FET. I am considering using a reed relay.

OK, use individual Nch and Pch FETs with a "L" suffix.

The "L" suffix means they can switch properly with a gate voltage as low as 4.5V.

It won't matter if they connect to Vcc, ground or somewhere in between.

 

Otherwise there are analog switches such as the Maxim 4514/4515 SPST with a resistance of only 20Ω at 5V

 

 

Thanks Oddbot, will look into that.

For anyone who is not aware of the original Rhex that inspired Markus, I highly recommend a visit to this page and this one and of course wikipedia.

The main properties that inspired me are:
- legs are compliant (they are springy): the compliance has been methodically tuned for their purpose;
- one motor per leg;
- one rotational, absolute encoder per leg;
- gait (leg synchronization) controlled by central brain;
- gait dynamically changes with circumstances and mission;

I'm speculating that a different kind of sensor (flex instead of rotation) would be advantageous to the energy efficiency of the gait. But if it would be easy (or even doable), I'm not sure.

I think you would be better off with using stepper motors. It seems to me that speed is crucial to get the foot back into position for the next ‘step’.
With stepper motors you have absolute control of speed and position. I would take it even further and see if there is a way to link three ‘feet’ with timing belt. This would give you absolute synchronization of three by three feet running from two steppers.