Pretty much summed up a lot of stuff I was working on for the last two years. With all the pitfalls I took before learning how to do things right. Loved the talk. Thank you.
Sensors should also take the AI-Actor actions in consideration. A sprinting or dodging enemy will not have much peripheral vision. Also: detection should depend more on a sum of observations over time, such that a clear line of light close by will ADD a lot to the sum, and a peripheral or obstructed target will add a smaller sum. The detection triggers when the sum get over a threshold.
Am I the only one watching this thinking this should be common sense? I feel like if you’ve simply played any metal gear solid game (before 5) then you’ve seen all this implemented with absolute perfection
It feels like a common approach but also it does not hit the nail on the head. The way he uses the Ray cast is a bit expensive and there are a lot better methods in doing so and most just require a bit more programming to get the AI just right but its better then shotting out a ray cast to see what is there.
This was a really good talk. Some really neat tricks.
Pretty much summed up a lot of stuff I was working on for the last two years. With all the pitfalls I took before learning how to do things right.
Loved the talk. Thank you.
Sensors should also take the AI-Actor actions in consideration. A sprinting or dodging enemy will not have much peripheral vision.
Also: detection should depend more on a sum of observations over time, such that a clear line of light close by will ADD a lot to the sum, and a peripheral or obstructed target will add a smaller sum. The detection triggers when the sum get over a threshold.
Really great talk! I just wish it was longer, with more time for questions
Am I the only one watching this thinking this should be common sense? I feel like if you’ve simply played any metal gear solid game (before 5) then you’ve seen all this implemented with absolute perfection
MGS even made most of it visible to the player, so it is indeed a good reference on how detection can work.
@@RSProduxx very true fam. Very true
It feels like a common approach but also it does not hit the nail on the head. The way he uses the Ray cast is a bit expensive and there are a lot better methods in doing so and most just require a bit more programming to get the AI just right but its better then shotting out a ray cast to see what is there.
14:00 (ignore this comment, it's just my personal timestamp)