Whee! Slaughterbots!
I know zero about this, but is that rigid reconstruct-localize-navigate-detect-follow flow really the way to do it? It feels overstructured.
It feels to me like if you asked me to walk into a room and track a person, I wouldn't try to map the whole area first. I'd just search for the person and dynamically avoid obstacles that appeared while heading for them. Sure, I'd build a mental map of the space, but that would be a continuous background thing; I'd start moving way before I had a full map.
The alternative is to let the AIs take actions like "go forward 3m", "rotate 25 degrees", etc at every iteration. But we found that 1) this makes the movement way too slow to follow anyone and 2) models are really bad at this.
Frankly, having an eval for something is the first step to get really good at it.
What if your eval finds out that all the Chinese models lag far behind the American models when it comes to drones? Or vice versa?
Also, the first step to do anything defensive about it. Evals are for sure dual use, but I'm leaning towards better to know than to stick your head in the sand.
Personally, I'm more worried about loss of control than misuse, which in my mind makes Chinese vs American capabilities less relevant (please enlighten me if I'm missing something here tho, maybe I misunderstood your question).
The point was that in that scenario the lagging side might make a deliberate effort to catch up turning something that was on nobody's mind into a race.
Thanks for the work! The AI-2027 forecast had the authors predict the AIs' abilities in robotics. Maybe the benchmark would help us understand more about this type of capabilities?
Hey everyone!
We’re introducing Drone-Bench, a benchmark where AI agents code drones to complete a simple autonomous surveillance task. Drone-Bench is independent but based on Project Pilot, our work with Anthropic.
I would make a condensed version here for LW, but visuals are much better on web: https://andonlabs.com/evals/drone-bench
Very curious on your feedback!