Not having a wrist is a real practical problem at times. Having a stiff wrist is better than none at all. But really, the wrist usually accompanies other actions (rather than leading actions)  and so, wrist motion has to be one that follows.

There are current papers detailing wrist control, such as voluntary wrist control. Now, given modern day prosthetics and their constructors at times lacking what I personally feel is the "proper angle" [e.g., 1,2], I seriously suggest to use some other type of control approach. You read it here first.

See, our brain has a tennis court sized area dedicated to controlling hand and wrist motion. Once that is unplugged, no other body parts can step in. If you were to replace that type of control with the rest of your body and there are people that attempt just that, you will need to employ forearm stump, upper arm, shoulder area, if possible you want targeted muscle reinnervation, and maybe you will be using the skin of your back, and then there are people that use foot switches. In other words, trying to do everything your wrist and hand did wilfully, you will run out of bodily real estate faster than you can say "blips".

Some folks go the hard way, all the way, and they like it like that. Currently there are ideas to use the thin bones of forearm stumps to drill even thinner bolts into them, then implant nerve interfaces and then use that to control prosthetic hands. I do wonder where they take all the energy required for such. And seeing as if our lives abound with prosthetic add-ons and whatnots, it is puzzling to see how paradigms of prosthetic hand usages are coined.

Obviously we are witnessing an odyssey. Regardless of the errors being made  there, I am a real fan of keeping things simple. I find there are concessions to be made. I also believe there are technical solutions to real problems to be found. And given the fact we still see no more than two electrodes for "bionic" hands - one for open, one for close - can we agree that there are no electrodes left to really control a wrist. That is, control a wrist willfully.

Controlling a wrist does not have to be performed willfully. I suggest to use any type of measurement devices to control a wrist unit in a way that stabilizes a prosthetic hand or gripper in angle (but not position) regarding 3D space without direct willful interference by the user.

That means, the prosthetic hand or gripper will be kept at its angle regarding absolute surrounding 3D coordinates (as absolute as these coordinates ever are, given humans on a planet earth) but not its position in that the hand or gripper is being moved around being attached to a prosthetic wrist and socket.

Here are the details.

1) Problem: typing.

When typing, I move the forearm across the keyboard. However, the finger that is extended must always point to the keys. The problem is that if the wrist is not there, other fingers can get in the way and typing becomes awkward. There are in fact videos showing tense shoulders and body posture when typing with a stiff wrist.

Using a hook, that problem is bypassed. I can maintain a relaxed shoulder and elbow position sińce a hook compensates for wrist shortcoming with its peculiar shape and angle. In fact, slanted hooks - with one gripper curving out a bit - both work to drink from a glass at an oblique angle and, at the same time, type a d grip at a rather straight angle. Wrist compensation becomes interesting mostly on conjunction with "bionic" hands.

2) Problem: balancing a glass or plate. 

When I hold a plate, and when I then walk around, the hand must stay at level always so as not to spill the plate. Same for drinks. Plates and drinks always must stay upright.

3) Problem: hammering and such. 

When hammering, the hammer should be guided with a stationary angle while the wrist compensates for forearm angles. Same for carving tools, saws, and other tools used with repetitive motion. This then also includes activities such as scrubbing, vacuum cleaning or ironing.

Clear? You may need an illustration for that : )

4) Problem: carrying baskets. 

When I swing my arm back and forth carrying objects, I want a well controlled wrist taking action for the basket to stay horizontal.

5) Problem: holding a tablet or smartphone

Holding a tablet computer or smartphone is greatly improved by having the device stay at the same angle regardless of the forearm or prosthetic socket position at any given moment. I could be sitting on a sofa and change position and still, I'd be looking at the device screen at the same angle during all the time.

6) Problem: control. 

All that is required for the user is a simple on/off switch. Switching the wrist stabilizer on will make the hand or gripper keep its angle from that moment on, switching it off will surrender the wrist to other controls.

Requirements:

Of course all these applications base on the clear expectation that any socket and suspension problems have been solved for myoelectric arms by now. Considering the sums spent for research tagged "prosthetic arms", by now, that should be a given.

References:

[1] Stiffness and position control of a prosthetic wrist by means of an EMG interface. Rao S, Carloni R, Stramigioli S. Conf Proc IEEE Eng Med Biol Soc. 2010;2010:495-8. doi: 10.1109/IEMBS.2010.5627153.
PMID: 21096538 http://ieeexplore.ieee.org/xpl/articleDetails.jsp?arnumber=5627153
[2] Development of a model osseo-magnetic link for intuitive rotational control of upper-limb prostheses. Rouse EJ, Nahlik DC, Peshkin MA, Kuiken TA. IEEE Trans Neural Syst Rehabil Eng. 2011 Apr;19(2):213-20. doi: 10.1109/TNSRE.2010.2102365. Epub 2010 Dec 30. PMID: 21193382

Update April 20, 2017

It took researchers 5 years to pick that idea up.

In the true sense of academic amputation research acknowledging what they see as prior works, my source (here) was not cited - but then, citing an amputee as that researcher that sets out to "turn arm amputees back into humans" necessarily requires them to see arm amputees as less human.

And yet, it is astonishing to not have them wait for 10, 30 or 50 years. So indeed, a paper by an unknown, certainly practically irrelevant but seemingly hopeful interested American researcher, Peter Kybert, now "officially" suggests the same as outlined above, and glancing over it, there appear to be photos of some testing [1].

kyberd2017accelerometers
.

Still, way to go.

[1] P. Kyberd and A. Poulton, "Use of accelerometers in the control of practical prosthetic arms," IEEE Transactions on Neural Systems and Rehabilitation Engineering, vol. PP, iss. 99, pp. 1-1, 2017.
[Bibtex]

@ARTICLE{kyberd2017accelerometers,
author={P. Kyberd and A. Poulton},
journal={IEEE Transactions on Neural Systems and Rehabilitation Engineering},
title={Use of accelerometers in the control of practical prosthetic arms},
year={2017},
volume={PP},
number={99},
pages={1-1},
keywords={Acceleration;Accelerometers;Elbow;Gravity;Motion segmentation;Prosthetics;Sensors;Accelerometers;Artificial Limbs;Prosthetic hand;Prosthetic limbs},
doi={10.1109/TNSRE.2017.2693683},
ISSN={1534-4320},
month={},}