Dynamic obstacles in Unity 5 free?

I am wondering whether Unity 5 will have dynamic obstacles included in a free version. Any ideas?

 

Unity hasn't decided yet which features come to Unity Free, this was mentioned several times in the forums and at Unite 2014.
I'm sure they will post it somewhere on the page, blog or forums when they are done.

 

I just hope at the very least the free features stay the same. And more Pro in Free is always welcome. xD

 

I don't think they'll make some free features pro-only. Remember when shadows were pro-only?

 

Yeah I was so happy when we got dat one dynamic directional light shadow. xD

 

Pardon my ignorance, what is this dynamic obstacle? Is it dynamic obstacle in endless runner type game?

 

Dynamic Obstacle Avoidance, for AI to navigate around moving objects.

 

In the current Unity version, navmesh can only bake still obstacles. If these objects move at runtime, agent will simply "ignore" its presence and will clip through it while avoiding the place where the object was originally located.

Realtime dynamic obstacles mean that moving objects such as chairs, tables are avoided by agents no matter where they are.

It's like lightmapping and realtime shadows.

 

Hmm, I suppose its application in an mmo where level environment meshes have animations or they move via coding. And then AI agent avoid them in runtime.

But can we not do this with coding? Say for example, dividing the whole area into grids, then raycasting or using trigger to detect those obstacle, then avoiding them on runtime? Haven't implemented this kind of thing before, just asking if this can be done or not.

 

It's quite expensive to use raycasts and triggers, especially if you're aiming for MMO since there are tons of AI.

 

Yeah, you definitely can do it yourself. The issue is that doing it yourself means either ugly hacks, or (re)doing stuff that the built in system already does just so you can add to it.

 

In the end I just resorted to Aaron Grunbergs A* Pathfinding for the AI.

 

I have a rough idea about it. Put your thought.
1. Get a list of AI agent which are alone on their respective grid.--this operation will be multithreaded on gpu via opencl or compute shader karnel

2. With gpu threadID, process each lonely agent in parallel for obstacle

Again being said, I have never used CUDA or openCL or directCompute along with unity before. So I am not sure of how things will go. But if it can be done, then with shear number gpu cores, we should not worry about raycast-trigger performance hit.