MX Simulator

Hey!

I had an idea about how AI bikes could improve on custom maps: Track creators having the possibility to “record” a lap on the track. One for all the different difficulties. Is this something that is possible to implement? or is it not possible?

Luutii wrote:Hey!

I had an idea about how AI bikes could improve on custom maps: Track creators having the possibility to “record” a lap on the track. One for all the different difficulties. Is this something that is possible to implement? or is it not possible?

Pretty sure it's already a thing. I remember doing it a long time ago, but it definitely doesn't work that good

Probably necessary to make them cheat to be really competitive. Really hard to make a good autonomous motocross rider.

jlv wrote:Probably necessary to make them cheat to be really competitive. Really hard to make a good autonomous motocross rider.

I know nothing about game programming so forgive me, but is there anyway you could apply machine learning to this sort of thing??

D_ROY178 wrote:

jlv wrote:Probably necessary to make them cheat to be really competitive. Really hard to make a good autonomous motocross rider.

I know nothing about game programming so forgive me, but is there anyway you could apply machine learning to this sort of thing??

I'm not an expert on it but I think the answer is yes. Pretty well defined inputs, outputs and goal. It'd be very computationally expensive though. The network would have to be really big since it would have to know about all the different types of terrain it could encounter. So I guess the actual answer is theoretically yes, practically no.

Watch this, then keep in mind it's a few pivot points and straight forward. As jlv summarized really good, practically, no.