Need some mechanical engineering help(hydraulics)

[kunkmiester]

It's not one of my crazier ideas. It involves master/slave controlled robot arms though, and it would be nice to get it all hydraulic.

The larger system arm would follow the movements of a linkage strapped to the operator's arm. Two ways to do it--have a hydraulic system that amplifies the master cylinders into the much larger cylinders on the arm, or have the master system all electronic, and encoders to help control the large arms.

To use electronics is a pain. Sure, industrial grade everything is available, but it'd have to be programmed and troubleshot, etc. Etc.

To do it purely hydraulically I'm quite sure can be done, but my Google fu is being weak. Master and slave bring up things like clutches and brake systems, and papers that seem peripherally related. The help I need is to find the terms so I can find and price the components. With that done I can actally think about telling people what this would cost.

Any help?

[hanelyp]

Operator moves the small arm, the large arm moves proportionately. Since you're unlikely to want force delivered by the large arm reduced in proportion to scale a direct hydraulic linkage between hydraulic cylinders wouldn't be satisfactory. So some kind of servomechanism is needed. These can be done with mechanically actuated valves. A differential gear set could serve to compare position readings on master and slave.

A nifty additional feature would be force on the large arm being reflected on the small arm. This might be had by a smaller hydraulic actuator on the master arm on the same output line driving the slave.

[ladajo]

How much range of motion do you need on the wearer's arm, and what is the relevant motion compared to the hydraulic arm?
Are you integrated within the working arm? or is this a remote application?
Are you intending amplification of force, and how do you envision feedback?

For google: try hydraulic haptic interface or haptic hydraulic interface.
I have not searched it, but that should give you some sort of useful results.

[krenshala]

Just because its interesting to me, I searched on "haptic interface hydraulic" and the first result seems relevant, if not exactly what you are looking for:
HAPTIC CONTROL OF HYDRAULIC MACHINERY USING PROPORTIONAL VALVES

[Diogenes]

Trying to build a Gundam, are you?



Not quite the same thing, but these guys have done something similar.






http://time.com/4003407/giant-robot-fig ... ots-japan/

[kunkmiester]

This was somewhat inspired by them, though I'm looking at a much smaller machine based on a skid steer. I'm seeing industrial applications too. A straight hydraulic system I'm guessing would be more familiar for troubleshooting.

The slave arm is on the same platform, and would be much larger. I'm guessing the master system would drive some sort of boost mechanism to get the power needed and you do want feedback.

That paper is quite relevant, I will have to peruse it later.

[krenshala]

Trying to build a Gundam, are you?

My first thought was a Landmate from Shirow Masamune's Appleseed series, since most models in the stories have the operator's arms visible below and the arms of the suit (image showing the layout of a landmate).

[DeltaV]

To do it purely hydraulically I'm quite sure can be done, but my Google fu is being weak. Master and slave bring up things like clutches and brake systems, and papers that seem peripherally related.

For what it's worth, I once had a gig improving a simulation of a high-precision fuel metering system. Control signals moved a small electric actuator, which moved a small spool valve (tiny displacements), which directed pressure to either side of a much larger fueldraulic actuator. I can't say more about the design without getting into proprietary data, but I can tell you that a simulation of such a system involves close attention to nonlinear spool valve characteristics near the zero position, hysteretic friction (best model: "elastoplastic"), exact mechanism kinematics, an estimate of slop in linkages (deadzone/backlash) and elastic bulk modulus of the working fluid.

[hanelyp]

I was making a simplifying assumption of master and slave being geometrically similar, allowing each input on the master to directly control a single actuator on the slave. Without that you get some complicated kinematic calculations to translate between motion spaces.

[Tom Ligon]

I used to operate and maintain an MTS servohydraulic fatigue machine. The servovalve on the main cylinder was electro-hydraulic but the basic mechanism could easily be all hydraulic. The bandwidth on this system was phenomenal. Small signal frequency response under precision control was something above 30 Hz (faster than most electric servos used in UAVs and model aircraft). The machine could operate precisely in stroke, strain, or load control, with movement limited primarily by the hydraulic pump capacity.

The servovalve had a small electric solenoid, essentially a meter movement, that drove a tiny lever in the valve. That lever, in turn, operated a small pilot valve system. Oil passed by the pilot valve operated a small piston that operated the larger valve. You could substitute a direct mechanical actuation of the tiny lever if you wished.

The key is that the servovalve needs to be controlled by a differential between your master and slave system, which should be straightforward enough to design.

Me, I'd go electronic control unless there was some compelling reason, like EMP tolerance, for the all-hydraulic. This allows feedback from multiple sensor types, such as load cells in a manipulator. The electronics really are not that daunting, and all the parts are available off the shelf.