Over the years, Valve has made dozens of changes to the system-level software behind SteamVR. Most of them aren’t inherently interesting if you’re not a VR developer. Then there’s the latest update, which Valve says was prompted by a change in the “limits of what we thought was humanly possible for controller motion. ”
After looking at “tracking data from experts,” Valve says it had to increase the theoretical limits for how quickly a human can move in VR. In the comments, Valve developer Ben Jackson details how top-level players were sometimes overwhelming the “internal sanity checks” that make sure SteamVR’s lighthouse tracking system is working correctly.
“One of these checks relates to how fast we thought it was physically possible for someone to turn their wrist,” Jackson writes. “It turns out that a properly motivated human using a light-enough controller could go faster (3,600 degrees/sec!) than we thought.”
For some context, 3,600 degrees per second is the same as turning your hand from palm up to palm down in just 50 milliseconds (0.05 seconds). That’s about four flaps of a hummingbird’s wing or less than half an average blink of an eye. Apparently such wrist speeds aren’t only possible but necessary to hit those flying blocks in some of ‘s toughest songs.
Road to VR’s Ben Lang goes into even more detail on why such quick wrist flicks could mess with SteamVR’s tracking algorithm. While SteamVR’s external Lighthouse tracking stations can provide extremely accurate tracking for the position and angle of the handheld controllers, they do so relatively slowly—only about 100 times per second.
Between those ticks, an internal inertial measurement unit (IMU) provides a quicker approximation of rotation and movement on the order of 1,000 times per second. But that quick inertial measurement can be imprecise and “drift” from the controller’s actual position in the real world.
Smoothing that out often requires a prediction algorithm based on where the system expects your hand to go next. Making that smoothing work in a sensible way, in turn, requires certain assumptions based on human kinesiology—assumptions that have apparently been upended by top-level players. That’s an incredible achievement and also a reflection of just how early we are in our understanding of how virtual reality can and should work at the basic human-interface level.
33 with 25 posters participating