New Year Bugs

First things first: We wish you a very happy new year! Today we present you our latest Polarith AI patch v1.0.1 including a handful of smaller improvements. Also, we are happy to tell you that we have managed to fix some serious bugs. The implementation of our path structure will require some more time than we initially thought (as our quality standards are very high), but you can tell that we are keen to show you soon! Furthermore, we have overhauled the library (in view of the upcoming features) and prepared the improvement of the UI as well, so you can enjoy our product even more!

Enhancements

  • AIMEnvironment can now use Unity’s layer system to define perceived objects.Polarith AI v1.0.1 Layers
  • It is now possible to change the sensor’s planar orientation in AIMPlanarShaper.
  • AIMContext indicator now considers the object scale.
  • Eliminated a remaining GetComponent call in the main update loop which caused 500 KByte of GC in editor mode.
  • Added a hierarchical distance check for the boundary behaviours to improve both performance and precision. Agents now use a fast approximation to check whether an object is relevant, and if it is, a more expensive but precise method is applied.
  • The general performance was improved to a noticeable degree.
  • Introduced new API methods for easily accessing epsilon constraints directly via AIMContext.

Fixes

  • The AIMPlanarInterpolator did not correctly consider if a target objective is to be minimized or maximized.
  • Using static game objects led to major problems when using behaviours which need visual bounds information.
    Polarith AI v1.0.1 Static Bug
  • Spread parameter in AIMPlanarSeekBounds and AIMPlanarFleeBounds did not work properly.
  • Eliminated an exception caused by the AIMContext indicator in the editor when using ShowReceptors.
  • Removed obsolete parameter (DistanceMapping) in AIMOrbit.
  • Fixed a bug in AIMContext leading to an unhandled exception while the minus button was pushed and there were no objectives.

What’s Next?

We are eagerly working on our very own inbuilt path structure (including a path following behaviour which can be used, e.g., for patrols) and on a patch improving usability a lot (which will polish the AI to a new level of efficiency, experience, look and feel).

We are very proud to share this information with you. As always, we are looking forward to your feedback and the work you are going to put out using our technology!