RVO Issue

What happen with RVO? My Example scene turn out to be like this…

The creators saya “all agent will fall back to not avoiding”, i dont know what it mean.

anyone know how to fix this issue? when i play RVO example, the agent going through each other. please i need this feature…


“all agent will fall back to not avoiding” basically means they will just pass through each other.

Unfortunately I cannot provide an ETA, and the future for this feature does look rather grim.

I sent you a mail about licensing, Aron. Would appreciate if you can take a look at my questions in the mail and hopefully you have some answers :slight_smile:

On another note; what is the plan going forward with regards to RVO/Local Avoidance in the A* Pathfinding Project? Are you planning to implement a new system? And what exactly is the issue with licensing? Is it somehow a problem because you’re using their system in an asset that you’ll get revenue off of? Or are they just not willing to license the RVO system at all?


I wrote a mail to UNC asking about if they would give a license to Aron and they answered today, that they are thinking about possibilities, cause several developers have asked about that.

Maybe theres a chance to get this feature again. And - just my opinion - RVO is worth some bucks.


It might be be a good strategy to ask other people to email UNC directly : )

(@mortennost, I believe I answered all your questions in your email, sorry for taking such a long time)

I really hope you get this sorted, but to shield yourself from more legal headaches you may want to remove the bit on your site and produce details about RVO as a feature. As I purchased this with the hope of using RVO at some point and if you are selling this as a feature but you cannot support it then its misleading to buyers…

Don’t get me wrong the product is brilliant even without RVO but I would say you need to stop touting it until you can support it again.

When I removed RVO, I thought it would be a short period until I could get it back. Now it doesn’t seem like that. I have added the large rvo notice and made it clear in the Free vs Pro comparison. But I guess I will have to remove it from the features list until I can get it back.

Why don’t you start working for Unity, Aron? I’m serious.

I’m sure they are jealous about your pathfinding solution. They seem to use also RVO.

Well, they did ask me (twice actually). However they were looking for a full time position, which I unfortunately wasn’t available for.

You’re kidding…

Aron … rethink your choice, man! Everyone would profit from you working for unity. Your pathfinding code belongs to the best ones available and you could get far more horsepower for implementing more features in a bigger team.

No kidding :stuck_out_tongue:

Anyway. The UNC people emailed me. Unfortunately their pricing is a bit steep.

We are planning to offer nonexclusive licenses to interested developers, providing each the ability to commercialize RVO2 in a single product (across multiple platforms). Requests to use RVO2 in subsequent products can be negotiated via future licenses or we can discuss building that into the initial license for an increased fee. For an initial, single product, we would offer the following options to prospective licensees:

(1) A one-time, up-front fee of $50,000 for use in a single product.

(2) An up-front fee of $500 for use a single product with milestone payments based upon the number of downloads/sales/etc. for that product ($5,000 after 100,000 units, $15,000 after 500,000 units, and $60,000 after 1 million units.)

Here a “developer” would not mean me selling the A* Pathfinding Project, but each user who uses it in a game.

Yeah, just got the same email today after talking to them a couple of weeks ago.
Is it just me, or does $50,000 sound a tad bit ridiculous.

The technology of the RVO2 library is not that advanced that, at least in my eyes, it justifies such an amount of money. At least not for a single product license…
Secondly, AAA-businesses with a lot of staff could probably be interested as the price is somewhat OK in terms of the expenses it would take to develop their own Local Avoidance technology. However, for indies this price is just too far out of reach…

Of course, the UNC guys have a copyright/trademark for the technology, so they can ask any price they want.

Still, I think the price was a tad bit ludicrous…

Yeah. Especially since you can just use the RVO feature in Unity Pro for $1000-$2000.

I have unity pro. can it be used with your A* pro package to provide RVO alongside your recast graphs and configuration interface?


No, using the unity rvo would imply using the unity pathfinding.

So what’s your plan then, Aron, in terms of Local Avoidance and the future of the A* Pathfinding Project?

If the UNC guys don’t realize they’re overpricing their library, it’s of course not an option.
Are there any other libraries out there worth using?

Still really want to keep our implementation of your Pathfinding project, since it’s really easy to use, and it is currently the best implementation of pathfinding in Unity for sure.

Hey, Aron!
Really sorry to hear about these RVO licensing issues.
Especially since the main reason I bought A star was because it was a complete solution that included local avoidance. Stinks to save up 100 bucks and then this feature doesn’t work.
But I understand you already know the complaints and are going through grief yourself. I really hope you get this issue figured out soon.
Do you have any suggestions or resources I can use to write my own local avoidance for A star, or are we going to need to wait until you figure something out?
If you have to abandon RVO avoidance, are you going to write or come up with a new avoidance system for us to use?


I am going to email them and tell them that I think their price is too high, I doubt it will lead to any significant changes, but I guess it is always worth a try.

Writing a new local avoidance system is hard. I have written local avoidance systems before, but they were no way near as fast as the RVO2 system (they could handle maybe at most 30-40 agents, while the RVO2 system can handle thousands). I could take a look at the Detour open source project which is actually what unity uses under the hood, but it would still take a lot of time to write one, I cannot promise anything unfortunately.

@mrpooley92, I understand. If you feel you cannot use the project now, I can organise a refund for you.

There are several force based local avoidance libraries for unity. Force based systems are usually not as good as rvo based systems, but they can work good enough. One such system is UnitySteer, I haven’t checked out the progress on that library in a while, but I used it a few years ago and it worked reasonably well.

Maybe its a rather good idea to use the obstacle avoidance code from detour?