Polarith AI (Free/Pro) | Movement, Pathfinding, Steering

Download Free | Buy Pro | Open-Source Formations

Fast and Easy to Use

Polarith AI offers all you need for achieving state-of-the-art movement AI with just a few clicks. Design complex movement behaviours and create immersive games or astonishing simulations by taking advantage of the sophisticated workflow.

Create, Prototype and Succeed

Our Unity plugin allows you to combine behaviours as easy as blending layers like in image manipulation tools. Every shipped AI behaviour can be parametrized easily to achieve completely different effects without even touching the code at all. Besides that, the system allows you to easily script your own behaviours with just a few lines of code.

The Context Matters

The major difference of our technology compared to classic steering approaches is that decision-making is based on the contextual information of an agent’s environment instead of simply weighting or prioritizing steering behaviours.

Much More Natural Decisions

You will massively reduce your implementation effort by using Polarith AI. Instead of concentrating on getting your character to move at all, you can focus on the design of concrete behaviours for creating an interesting movement that makes your project immersive.

Features

Free & Pro

  • Easy to use with one-click setup
  • Design-oriented and reusable
  • Highly performant and scalable
  • Over 20 combinable behaviours
  • Active/passive collision avoidance
  • Dynamic obstacle avoidance
  • Single, crowd and swarm simulation
  • Local decision-making
  • Context steering
  • Line and circle sensors for 2D/3D
  • Extensible C# API

Pro only

  • Unity pathfinding integration

  • Customizable paths and patrols

  • Collider and visual bounds

  • Aligned and oriented bounding boxes

  • Unity NavMesh integration

  • Fully editable 2D sensor shapes

  • Level of detail AI technology

  • Multithreading with load balancing

  • Prioritized support guaranteed

  • No Polarith logo during app start

Upcoming

  • Spherical 3D sensors

  • A* Pathfinding Project integration

  • Component master

  • Formations

  • …

Get Started with Tutorials and the Documentation

More Videos…

Quick Start | Manual | Component Reference | API Reference

FAQ

Does the plugin support pathfinding?
Yes. We support custom path structures (for patrols, racing tracks and so forth), dynamic waypoint following and NavMesh-based pathfinding which is fully integrated with our steering algorithms.

Is local avoidance supported?
Yes. The approach we use is called context steering. Based on this technique, local avoidance is achieved implicitly as well as explicitly. If you want to go into detail, please, read the documentation.

Can I do the AI setup at runtime?
Yes. Everything that you can do with Polarith AI in the editor (all this handy drag and drop and presetting) can easily be done via the scripting API at runtime, too. The API offers you even more possibilities than the inspector does.

Can I add custom behaviours?
Yes. You can add as many custom behaviours as you like via C# code. Hereto, we provide a tutorial in the documentation.

Does Polarith AI support 2D and 3D applications?
Yes. Optionally, your game objects can be projected onto a plane for both Unity modes 2D and 3D.

Do you support context steering with spherical 3D sensors?
Not yet, but we are currently working hard to integrate full mesh-based 3D sensors.

How can game object data be provided for the system?
You can set them directly or take advantage of our perception pipeline, whereby Unity’s layer mechanism is supported. By using Unity’s layer system, you can easily group different kinds of game objects and improve the overall system performance.

Made with :heart: in Germany

12 Likes

This is great that you’re now on the forums.

I do look forward to new tutorials, especially seeing examples of behaviours expressed via both the UI and API.

1 Like

Great! This looks like a well-made and documented asset. Good thing there aren’t really any great specialized steering assets so far either.

1 Like

Hey, guys. Yeah, it was definitely about time to be here. Feels great! :slight_smile:

We are currently planning and working on the next tutorials we are going to release on YouTube. I think there will be a lot of interesting things for you to discover. Formations, flocking, integration into different kinds of game genres and a description on how you can combine our approach with pathfinding very easily. Moreover, we are also working on meaningful demos for platformers, racing and space games. So much to do!

Thank you @christoph_r . We spent a lot of time in system architecture and design to make the usage as easy as possible while providing the best performance. We are convinced that you guys will love it.

A big thank you to all supporters and early adopters. You rock :sunglasses:!

Martin from Polarith

1 Like

I look forward to the demos, especially any racing scenarios. Then again, any collection of (easily digestible) examples makes for an excellent starting point.

Looking forward to your 3D implementation. There aren’t many 3D steering solutions available currently.

I just find this today, and exciting to learn it.
Great Job!

Hey @unicoea , then let me welcome you to the Polarith family. Nice to have you here.

@Korindian : That’s great. We definitely want that feature as well. You might be happy to hear that one dev is currently all in programming this. :slight_smile:

Martin from Polarith

I remember taking a look at the Unity Steer (open source and free) as well as attempting to implement my own basic understanding of steering behaviors, and this one seems the most promising due to a great deal of interaction through tutorials and answering questions from other users. I am looking forward to toying around with it to find out more about implementing waypoints and such.

Hey @MrJBRPG , nice to have you here.

When the two patches are released we are currently working on, implementing waypoints will be nothing more than adding and parameterizing a component (either graphically by scene view or via API in-code).

You guys might be happy to hear that the patch we are going to release next week will make the current Polarith AI plugin become a Pro version (which will be more feature-rich and, hence, more expensive then). All of you which already have Polarith AI will become a Pro owner. Meanwhile, we will upload a second (light) version of the plugin which will be available for free (for all of you who want to test things instead of buying the cat in the bag ;)).

One week or two weeks later (depending on how fast the UAS admins are), we will release another patch adding a fully customizable path component including full pathfinding integration with NavMesh support to the Pro version. This will bring the best of both worlds together: Global movement solutions by pathfinding combined with local movement solutions by our context steering. So your AI can behave as natural as possible through context steering while being as precise as pathfinding can be.

If you are definitely interested in the stuff, consider buying the plugin right now as long as it is at its current price. It doesn’t get cheaper.

Martin from Polarith

2 Likes

I know this is AI for movement… but I am wondering… really it’s just AI for movement? Are there any other features besides the movement?

Hey @Cartoon-Mania ,

our plugin is all about multi-criteria optimization (MCO or MOO for multi-objective optimization) which we currently apply on movement directions (and magnitudes). Of course, the underlying framework (API) can be used to optimize any MCO problem where you can obtain objective values for.

For example, imagine some kind of droid. Now, if you derive your own data structure from our base classes and fill the objectives with values for goals like “maximize battery life”, “take no damage”, “travel the shortest distance” or “support your allies”, our MCO algorithms would find parameterizable good solutions for these criteria in real-time, too. So, our core algorithms are all about general decision-making. Because games are about animated/moving graphics, we decided to utilize these techniques first for realizing natural-appearing movement => AIM (AI for Movement).

With the support of you guys, it may come the day, we might be able to extend our workflow to support general decision-making as comfortable as the movement (maybe called AID then => AI for Decision-making :smile:). That always depends on how many developers can be financed by our work.

Martin from Polarith

Niiice. Was planning to do that myself, so having both integrated into one asset taking care of AI movement is sweet. I hope there will be fine-tuned control over which solution is going to be active when and how the two will interact.

AI for Decision-making is great. This is a dilemma I buy it if it is a decent making AI. I hope that day will come soon.

Is the droid you said possible in the current version?

@christoph_r : Of course, we release new features only when we love them ourselves (so in the end, we are our most annoying customers). :eyes: Fine-tuning via inspector will be intuitive and easy enough. Code-affine users will be able to control every bit per API as usual. In the case you don’t like what we do with the path, then tell us rudely here in this thread. That would dramatically increase the chance that we refactor it in the next patch. :smile:

@Cartoon-Mania : Possible: Yes, Comfortable: No. In short: If you only desire general decision-making and in terms of implementation effort, you would be better off with comfortable graphical tools based on fuzzy or utility logic. There are a couple of decent solutions in the store dealing with this task.

Martin from Polarith

1 Like

Woah… Now that is one promising teaser for upcoming features. Even when things come out nice and slowly, I am likely to try to find ways to create more sophisticated versions such as follow formation, waypoint array open and closed (like in racing tracks and such) while you guys are working on providing elegant solutions.

Also, it is awesome to finally have a all-in-one movement AI devs that are actually responsive.

Only one question:
I like to know about Order - if it is a higher number, does it take precedence over lower numbers or is it the other way around because the execution order is not yet clear.

The higher the number, the later it will be executed within the pipeline of behaviours writing objective values into the context maps.

@MrJBRPG , thank you for your feedback, we will consider this topic for updating the docs in the next patch.

Martin from Polarith

That explains price increase, I saw this a couple days ago but was looking/trying other assets and waited to today to buy it and notice the price had increased. Oh well at the current price I will have to wait to try it out before getting it.

Hey @Space-Odyssey , well, the Free version currently waits for pending review. It should be online during the next days depending on the Asset Store admins. We will announce this here when it can be downloaded.

Martin from Polarith