Prof. Dr. Patrick Baudisch

Mutual Human Actuation

Human actuation is the idea of using people to provide large-scale force feedback to users. The Haptic Turk system, for example, used four human actuators to lift and push a virtual reality user; TurkDeck used ten human actuators to place and animate props for a single user. While the experience of human actuators was decent, it was still inferior to the experience these people could have had, had they participated as a user. In this paper, we address this issue by making everyone a user. We introduce mutual human actuation, a version of human actuation that works without dedicated human actuators. The key idea is to run pairs of users at the same time and have them provide human actuation to each other. 


Our system, Mutual Turk, achieves this by (1) offering shared props through which users can exchange forces while obscuring the fact that there is a human on the other side, and (2) synchronizing the two users’ timelines such that their way of manipulating the shared props is consistent across both virtual worlds. We demonstrate mutual human actuation with an example experience in which users pilot kites though storms, tug fish out of ponds, are pummeled by hail, battle monsters, hop across chasms, push loaded carts, and ride in moving vehicles.

The user on the left, alone in his virtual world, is trying to pull a huge creature out of the water. He feels how the creature is struggling and pulling on his fishing rod. At the same time, the user on the right, also alone in her virtual world, is struggling to control her kite during a heavy storm, which is whipping her kite through the air. 

While users’ experiences of force might suggest the presence of a force feedback machine, Mutual Turk achieves force feedback instead using shared props that transmit forces between users. The system orchestrates users so as to actuate their prop at just the right moment and with just the right force to produce the correct experience for the other user.    

Shared Props

The key to enabling mutual human actuation is the use of shared props. Props allow users to exchange forces without revealing that these forces are generated by another human. If users’ hands or any other part of the users’ bodies were ever brought into immediate physical contact, this would instantaneously give away that the other side is human (through, e.g., skin softness, temperature, moisture, shape of hands). This is what the shared prop prevents. Thus, one way to think of Mutual Turk’s shared props is as a means of masking the information exchanged between the users.

Different prop designs enable different levels of expressivity. We explored five: continuous force (shown above), moving, impact, contactless sensations, and rearranging props (least expressive).

Continuous Motion

(a) the user pushes an empty cart under a faucet. (b) He watches as the faucet drops water into the tank. (c) Meanwhile, in the other user’s world, the world is collapsing and our user is making a dash for the escape pod. She hops onto the escape pod (thereby giving weight to the water tank in the first user’s world). (d) She then rides the automated pod down an evacuation tunnel—(propelled by the first user pushing his cart), (e) just as the first user starts to push his (now much heavier) cart on to next destination. The office chair prop transmits movement and rotation in one direction. In return, the user sitting on the chair affects the chair’s inertia. Unlike the interactions in the previous category, only one user’s hands are involved in driving the office chair. This allowed us to drop the tether and use a rigid prop instead.


(a) The user sees himself walking in stormy weather; he sees huge hailstones shooting down from the sky at an angle, hitting his body at various locations. (b) In the meanwhile, the other user is fighting back a monster using an improvised weapon made from a plastic tube she found at the lab. The foam stick used in this scene touches the other user for only very brief periods of time, which properly obfuscates the origin of the force.

Contactless Sensations

(a) The user is trying to fight her way back to lab against very heavy wind. (b) Meanwhile, the other user in the other world is trying to get a fire going to distill the emulsion created earlier.

Rearranging Props

(a) shows a user waiting for a series of pillars to rise in order to allow her to cross the pit ahead of her. As she lowers her right foot to probe the space below, she can feel the void. Once she sees that the pillar has fully risen, she can step on it. (b) In the meanwhile, the other user is solving a puzzle that requires him to place numbered boxes on matching tiles. This is the least expressive type of exchange between two users as no physical contact between the two users is ever established. It thus is also the most obfuscated type of interaction.

Synchronizing Users

Mutual Turk must not only synchronize the use of the shared props, but also their acquisition and disposal. A typical scene consists of a period of real walking within a designated area, the acquisition of a prop, the use of the prop forming an action sequence, the disposal of the prop, and return to unencumbered real walking. Experiences are then sequences of such scenes.

Contribution and Benefit

Our main contribution is the concept of mutual human actuation. The main benefit of this approach is that it eliminates the need for dedicated human actuators, instead allowing everyone to enjoy the experience of a user. At the same time, Mutual Turk still offers the benefits of human actuating system, i.e., it is allows creating human-scale force feedback without mechanical machines. We have created a proof-of-concept implementation.


Cheng, L., Marwecki, S., and Baudisch, P.
Mutual Human Actuation
In Proceedings of UIST'17. 
 PDF  Talk (youtube)