Dexta are Making Haptic Exoskeleton Gloves that Lets VR Push Back


Dexta Robotics is pioneering what they see as the next logical step for immersive experiences, being able to touch the virtual world, and to feel it push back. Their Dexmo VR gloves are designed to provide resistance to the wearer, corresponding to their interaction with the VR environment. In this guest post, Dexta Robotics founder Aler Gu brings us up to date with Dexmo’s progress.

aler-gu-headshot-dexmo-cropAler Gu is the founder and CEO of Dexta Robotics. He is a passionate roboticist and a researcher from University of Cambridge. His paper on hand VR interaction was published in the ACM CHI 2016 journal. Dexta Robotics holds a U.S. patent and four Chinese patents on their haptics technology. Dexta specializes in hand VR interaction and hand force feedback. Their main project is Dexmo, a consumer-friendly hand exoskeleton ideal for use in VR.

VR is now on everybody’s mind. Immersive technology that stimulates sight and sound are promising and progressing quickly. But touch is the key to immersion. To feel and interact with virtual objects is to reach across the digital divide.

True force feedback has been possible in a lab. But only with large, loaded motors. Cumbersome, expensive, untenable for mass markets. The idea behind Dexmo is to make force feedback light, affordable and workable.

The key is binary force feedback. Intelligently-controlled slider mechanisms in the exoskeleton that lock and unlock joints. This technique provides force feedback at the individual finger level in VR. The tech also captures hand motion in 3D space. The combination is satisfying and affordable. Portable and wireless.

What Any VR Game Can Learn From the 'Electronauts' Interface – Inside XR Design

Binary, in this context, refers to the two states of the force feedback unit (FFU) on each finger exoskeleton: on and off. When the FFU is in off state, users can move their fingers. Dexmo tracks their finger movement. Unlike other feedback devices, it doesn’t need the motor to take all the torque the user applies. That works to simulate absolute rigid force feedback but doesn’t work for springy feedback. (The middle part of the force output spectrum.) That’s why Dexta developed another Dexmo prototype with variable stiffness, to explore other possibilities.

Dexta has a granted and sealed US patent for the technology that makes this possible, and our research has been published in SIGCHI. The progress we’ve made is tangible, and we’re excited about it. The below video, which we’ve moved far beyond, provides a glimpse at what we’re trying to achieve and how our technology looks in use.

With this technology and knowledge, we’re refining the binary force feedback to include softness. This approach relies on careful impedance control stiffness adjustments. The video at the top of the article demonstrates this.

We control the motor to act as a spring. The stiffness of the “spring” adjusts electronically in real-time, controlled by the application. This control method is standard in collaborative robotics, a field focusing on human augmentation through robot apparatus. Japanese research teams had similar ideas, but they never pushed it to market. Dexta is the first to apply this to a commercialized product for hand exoskeletons.

How will this force feedback technology affect your VR experience? When inside the virtual environment, you can feel the difference between elastic and rigid virtual objects. You’ll be able to hold a gun and feel a realistic “clicky’ level of resistance from the trigger. More subtly, you’ll be able to pick up a virtual object and discern by touch whether it’s a rubber ball or a rock.

Hands-on: Logitech's MX Ink for Quest Could Legitimize VR Styluses as a Whole

VR content creators can dynamically change the stiffness of virtual objects through the use of our SDK. We have a video of an early prototype of a variable stiffness force feedback exoskeleton in action.

The video above was taken from some of our much earlier R&D work. We’ve come a long way since then. For example, we’ve managed to miniaturize the binary force feedback module on each finger, designing custom servos in the process.

We’re almost there. The new versions of our Dexmo glove will bring new possibilities to virtual reality. While ever evolving advances in VR visuals and audio bring us closer to the digital divide, Dexta is taking steps to help us reach across it, sooner rather than later.

This article may contain affiliate links. If you click an affiliate link and buy a product we may receive a small commission which helps support the publication. See here for more information.

  • wheeler

    Very cool, I would love to see this come to VR. I’ll be interested to see how this progresses.

    However, it would seem to me that once users become accustomed to this, at a minimum they will desire constraints at the wrist, elbow and shoulder.

    • RadthorDax

      Imagine the delicious confusion when they rest their hands on two VR surfaces either side and try to lift themselves up!

      • DougP

        Not funny – I’ve already gone off balance thinking for a moment to *lean* on a virtual desk to stand from a crouched position! :)

        Funny how much VR tricks the brain.

  • Firestorm185

    I remember seeing something about this about a year ago. Can’t wait for this control type to come to VR mainstream.

  • Jason

    I wonder if a compact design glove could be made using Magnetorheological fluids:

    In theory, you can have variable mechanical resistance to motion.

    I found a decent amount of hits for MR fuilds and haptics, but they were pretty much all industrial. There was one dealing with remote surgery which looked promising.

    With a glove that had this fluid in various channels, it could basically “freeze” in any position it needed to. I don’t believe it could actually change your finger position itself (which is good on account of not giving skynet a way to break your fingers), and pulling the power would cause the whole glove to relax (making it much less likely you’d get “stuck” in it).

    One downside is that it’s a bit pricey for the fluid:

    And according to wikipedia, “Fluids are subject to thickening after prolonged use and need replacing.” But it’s hard to tell what “prolonged use” would mean in a glove for gaming compared to in an industrial setting.

    I’m also not sure about how large the electromagnet would have to be or how much fluid you would have to use to give the proper amount of resistance.

    • temp

      That is a really nice idea! I’ve also heard about other substances in development that act the same way when current is applied, for example:

      Sadly they are not available right now afaik, but perhaps you could prototype such devices with the more expensive Magnetorheological fluids approach.

      In any case, exoskeletons are cheap and accessible; making them good for releasing to the general public to facilitate software support in preparation for less cumbersome and comfortable(?) haptic gloves that could reach a wider spread.

      • Jason

        Yeah, the MR stuff was the least hard to get of all the exotic stuff I looked into.

        I was also pondering artificial muscles made of coiled fishing line:

        Talk about low tech stuff you could build in your garage! You wrap a wire around it to heat it up (a much less bulky approach than an electromagnet) and you embed it in a tube that’s water cooled.

        I THINK that would work, but it’s hard to get any solid answers. I’m also skeptical of how you could mass produce such an item (even more so than the MR gloves). I also wonder how much you can vary the tension to get just the right level of force.

        But in theory, you could make a glove that had these artificial muscles that could put put force on the backs of finger joints to pull them back open. You’d put them inside of rods so that you couldn’t bend anyone’s fingers backwards or anything.

        Heck, at how cheap this tech is, you might could even make ones to go over your elbows, shoulders, knees, etc.

    • If you use the fluid as a brake, to slow or stop movement at a radial joint, you wouldn’t need more then a drop worth for each joint. They could even be in self contained packs that could be swapped out as needed.

      As you move, the fluid would be shunted between two tiny reservoirs. Apply the magnetic field in the valve between the reservoirs, and you can stop motion as needed. Each of these, again, would be like a droplet. The idea isn’t to strong-arm the hand, but just slow down the fingers to give some presences.

      Brakes on a car aren’t huge, they are just in a good location to stop the wheels. Follow that same design philosophy.

      • Jason

        Well, I was thinking of discs/rectangles at the joint, but I’m afraid if they weren’t connected joint-to-joint, you could still flex. Imagine taping a penny to the top of each joint and a pipe cleaner from penny to penny. The pennies wouldn’t stop you from extending your finger, they’d just cause the pipe cleaner to bend. Does that make sense? But depending on the strength per volume, I think you could probably get away with a little “cage” type contraption. A channel or two of MR fluid that extends along the whole glove and a few ring “anchors” (kind of like the rings in the video). The rings could probably just be made of strong fabric/plastic as they don’t need to stiffen.