Radiance - Dart port of gdx-ai

I wasn't able to use new version of flame_test, because it depends on Flutter SDK, and we want Radiance to be a pure-Dart project.

The interim solution is to simply copy the closeToVector method over from flame_test, and use that. In the future we might be able to use flame_test directly.

Added seperation feature, check it!

Recording :

https://user-images.githubusercontent.com/96647299/195695451-bffcef73-4d95-463e-a67c-a42bcb23cc1a.mp4

Thank you.

Closes #18

This is a somewhat preliminary implementation of the Seek behavior for two different types of Kinematics.

Examples/tests will be added later, when I figure out a system for them.

• Agents with LightKinematics will now always face the direction of their velocity.
• Seek behavior for LightKinematics agents will now stop at the target without changing the orientation.
• Reducing maxSpeed of a LightKinematics agent will now also scale down its velocity.
• Field Kinematics.own renamed into owner.
• Fixed angle calculations in Sandbox.
• Cloning of Kinematics objects removed (since it cannot be adequately expanded to cloning of behaviors).

• Added LightEntity (with LightKinematics)
• Added several more presets
• Scenes can now be explicitly saved/restored
• Removed cloning of entities

A few notes on the design of Scene class, concerning the ability to "reset" the scene to its initial state how it was at the beginning of a simulation:

• At first I thought we will have scene factories, i.e. a preset will not contain a scene but rather a function to create a new scene from scratch. This works fine, except when we want to have an "editable scene" functionality in the future;
• Then I wanted scenes to be clone-able, so that at the start of the simulation we would clone the currently selected preset, and run simulation on the clone. Unfortunately, cloning proved to be hard, especially for entities that have behaviors targeting other entities. The targets for behaviors are declared final so cannot be reassigned, and for a good reason: a behavior may need to perform some initialization work depending on the target;
• The final solution is to save the "state" of each entity at the start of a simulation, and then later revert to that state. So far I don't see any problems with this approach.

There is now a button in the top menu that toggles whether those vectors are shown or not.

In addition, sandbox Entitys and main library Kinematic objects can now be cloned.

Rename MaxSpeedKinematics -> LightKinematics and MaxAccelerationKinematics -> HeavyKinematics. This makes the names closer to the other kinematics classes that we plan to add in the future: PedalKinematics, CarKinematics, RobotKinematics.

This adds several control buttons at the top of the Sandbox widget: to start/pause/stop the simulation. More buttons can be added in the future as needed.

This is an initial implementation of the Sandbox -- eventually it should get many more features, including the ability to create and edit scenes.

The Sandbox is a bit like Examples book in the Flame repository, only here the ultimate goal is to allow the user to modify each scene as they see fit. In particular, this will allow the user to experiment with different settings, different mixtures of behaviors, etc.

Description of Separation from https://www.red3d.com/cwr/steer/gdc99/:

Separation steering behavior gives a character the ability to maintain a certain separation distance from others nearby. This can be used to prevent characters from crowding together. To compute steering for separation, first a search is made to find other characters within the specified neighborhood. This might be an exhaustive search of all characters in the simulated world, or might use some sort of spatial partitioning or caching scheme to limit the search to local characters. For each nearby character, a repulsive force is computed by subtracting the positions of our character and the nearby character, normalizing, and then applying a 1/r weighting. (That is, the position offset vector is scaled by 1/r 2.) Note that 1/r is just a setting that has worked well, not a fundamental value. These repulsive forces for each nearby character are summed together to produce the overall steering force. See Figure 14: