Why is there still no truly omnidirectional treadmill? (technical reasons)

Question

Disclaimer: I'm not a (device) engineer. I don't know the terminology. So I suppose some of my statements, questions or ideas could sound either naive or vague.

---

As far as I undestand, the most advanced (and publicly known) omnidirectional treadmill is the Infinadeck. But with current state of the progress on this treadmill, it's clearly not ready for a mass market. One of my biggest concerns is the shape of the treadmill: why don't the Infinadeck's engineers use a disk instead of a square?

I was thinking about possible implementation. I'm not sure how easy it would be to implement. Because, of course, the biggest issues are often related to implementation details. I would like to hear your thoughts about it. As well as some comments on Infinadeck's shape.

This description isn't touching the vector generation or movement initiation, it's just about directionality of the treadmill.

---

Figure 1.

---

Figure 2. Single engine: the sphere and the conveyor parts. This figure doesn't show the spikes, though.

---

So the shape of the device is cylindrical. There are three major elements: shell, engines and platform.

1. Shell is just a container for the engines.
2. Platform is a thick metal disk wrapped in a thin elastic and seamless case. So it's like in regular one-directional threadmills, but radial. The case, of course, is smooth on the inside to lower the friction. But on the outside it has numerous tiny sockets that are used for pulling the case in the desired direction. And of course it's not fixed by some kind of joints or anything. It lies on top of the engines. There also may be some supporting columns beside the engines. On the other hand it needs to be restricred by the shell, so that it stays inside.
3. There are multiple engines, spread inside the shell. Each one of them consists of the two major parts:
   1. The sphere that translates its movement into movement of the case above it. The sphere isn't a smooth ball. It's covered with numerous spikes all around it which precisely fit the sockets of the platform's case. So when the sphere rotates, it pulls the case in the same direction.
   2. The same spikes are used to rotate the sphere. There is a pair of conveyor belts(see Figure 2) that translate their movement into the sphere's. I suppose the easiest way is to make the spikes penetrate the belts. And this pair of conveyors can move around the sphere horizontally depending on the required direction but independently of the vertical movement of the belts itself.

---

UPD: The platform should be able to move in any direction at any moment (with reasonable latency, of course). And the vector is generated in real time by Neural Network watching the user of the treadmill. So the person just walks in random direction and the neural network instantly generates the vector, which is then translated into engines movement.

---

How far this concept from reality?

Answer 1

A disk shaped threadmill seems somewhat problematic from a technical perspective. Placement of actuators and moving of the surface material seems much more problematic in a disk than other shapes. Speed of actuators need to be nonuniform for one.

Tension on a disk seems like a unknown problem. On a disk the corners make up more of the area while in a square its uniform. Wether you can even roll such a surface at all without elevating the user quite much is a question i cant answer. In anycase a disk shaped structure needs to withstand much more two directional bending and nonuniform stretching.

But really many of the technical problems could be solved by adequate funding.