just found something in the interwebs, as the title suggests its a paper on a collision avoidance sheme for scenarios with many independent actors.
seems to be pretty performant in that benchmarks they claim to have run.
http://pcl.intel-research.net/publicati ... ca2009.pdf
Post
Thu Jan 07, 2016 8:30 pm
#2

If you're trying to fill the multiverse, and you're not willing to consider the entire creative output of humanity as a starting point, you're wasting your time.
Re: Collision Avoidance for Multi-Agent Simulation
Sounds neat... If you have 64 cores to work with 


If you're trying to fill the multiverse, and you're not willing to consider the entire creative output of humanity as a starting point, you're wasting your time.
Post
Fri Jan 08, 2016 1:35 am
#3
geneous agents".
We dont have hundreds of thousands of agents
Re: Collision Avoidance for Multi-Agent Simulation
64 cores for simulations "with few hundreds of thousands of hetero-Hyperion wrote:Sounds neat... If you have 64 cores to work with
geneous agents".
We dont have hundreds of thousands of agents

Post
Fri Jan 08, 2016 5:19 pm
#4

If you're trying to fill the multiverse, and you're not willing to consider the entire creative output of humanity as a starting point, you're wasting your time.
Re: Collision Avoidance for Multi-Agent Simulation
A goal for LT 2 then. 


If you're trying to fill the multiverse, and you're not willing to consider the entire creative output of humanity as a starting point, you're wasting your time.
Post
Sat Jan 09, 2016 2:47 am
#5
Re: Collision Avoidance for Multi-Agent Simulation
I'd be happy with 500 NPCs flying around in densely populated systems. I'm sure that Josh is stretching the LT engine to its max so that we get very smart AI as well as lots of them.
