There's an interesting relationship between what developer Embark Studios is doing with new updates, like the recent Arc Raiders Flashpoint update, and what it can feasibly do with the technology available to it. In the future, as machine learning research lead Martin Singh-Blom told me at GDC, Embark will surely develop new tools and techniques that may enable ideas that weren't possible in the past – for instance, rope physics sophisticated enough to support attacking large Arc with tripwires.
"I think we all have different ideas," Singh-Blom says. "Those of us who really work with the physics and make things happen with physics of course want more physics in the game at every turn. We want ropes to pull at legs and wires to trip them, like the Hoth invasion scene in Star Wars where they have the speeders and wrap up the legs. We could do that if we just had better rope physics, right?"
Singh-Blom throws the example out casually – and he's talking about using tripwires, not flying around in speeders, for the record – but it's a compelling idea. Imagine slowing or briefly stunning Queens or Matriarchs, or temporarily disabling Leapers, by using some sort of tripwire emplacement at just the right angle. But there's a difference between dreaming this stuff up and implementing it, not just at all but in a practical and consistent way, and that's where ongoing R&D comes into play.
"Those kinds of things, we keep pushing for," Singh-Blom says, "and then the designers are like, 'Well, it doesn't fit in our game right now. This is not the vision for this part, or we just don't have the resources to deal with this kind of stuff right now.' They tend to be much more hooked into what the game actually needs, and we tend to be much more hooked into what we want to do with this new tech. Hopefully we get a good synthesis where we push the boundaries of what we can do, and still do things that make sense for the game."
Singh-Blom offers another example: resurrecting Arc ideas which were cut during development, sometimes because they didn't fit the game, and sometimes because they would be too difficult to complete. Another case where technological developments could inform design is in dampers, a way to refine and essentially round out Arc leg movements by adding shock absorption to their legs, almost, to my mind, like suspension on a car.
"I think we could get much nicer, smoother gaits, but when we try that, the physics don't like it and things get wobbly and the simulation gets bad, and it's very hard for the robots to look good in those settings for some reason," he says. There's also the resource cost of features like this to account for: anything Embark adds in will eat something server-side, which can add to overhead costs. Even with an Arc Raiders-sized warchest, Singh-Blom stresses that Embark can't just do anything it wants in this space.
"It's kind of fun to see the kinds of things that we in the research team are constantly pushing for," he concludes. "New capabilities in robots of all kinds. Once we get a breakthrough for one robot, it usually translates really well to the other robots as well. Things like physics upgrades or methodological breakthroughs for our research team lead to new behaviors."
