Steer behaviors allows you to control the locomotion of “characters”, this can reduce drastically the time needed to develop a game since for almost every game we need to set how these “characters” will be moving around the scene.

YouTube
Steer behaviors in action

youtu.be/yyzTntsGV00

Be sure that you have checked the demos: They can be downloaded here: jmesteer.bdevel.org

First steps

The steer behaviors AI is integrated with MonkeyBrains so before start coding be sure that you have checked the monkey brains documentation.

Be sure to create a reference to MonkeyBrains from your project.

Finally, do not forget to import the com.jme3.ai.agents.behaviors.npc.steering package.

Overview

Avaliable behaviours:

  • Move

  • Seek

  • Arrive

  • Flee

  • Pursuit

  • Leader follow

  • Evade

  • Cohesion

  • Alignment

  • Obstacle Avoidance

  • Unaligned obstacle avoidance

  • Hide

  • Slow

  • Queuing

  • Containment

  • Path follow

  • Wall approach

  • Wander Area

  • Simple Wander

  • Relative wander

  • Sphere wander

  • Box explore

  • Separation

All the behaviours extend from the AbstractSteeringBehavior class.

Adding a steer behavior

Create instances from the steer behavior classes, They are located in the com.jme3.ai.agents.behaviors.npc.steering package.

If we want to add more than one steer behavior, we need to create a container:

If you add more than one steer behavior to a SimpleMainBehavior it will cause problems in the rotation of the agents.

ContainerPurpose

CompoundSteeringBehavior

Contains and merges several AbstractSteeringBehavior instances

BalancedCompoundSteeringBehavior

Each force generated inside this container is reduced in relation with a proportion factor: “Partial Force / “Total container force

Once we know which container fits better for our agent, We create a new instance and add all the behaviors that we need:

SimpleMainBehaviour mainBehavior = new SimpleMainBehavior(myAgent);
    CompoundSteeringBehavior steer = new CompoundSteeringBehavior(myAgent);
    //BalancedCompoundSteeringBehavior steer = new BalancedCompoundSteeringBehavior(myAgent);
    steer.addSteerBehavior(steerBehavior1);
    steer.addSteerBehavior(steerBehavior2);
mainBehaviour.addBehavior(steer);
myAgent.setMainBehavior(mainBehavior);

Note that you can have nested containers, like is shown in the following example:

SimpleMainBehaviour mainBehavior = new SimpleMainBehavior(myAgent);
    CompoundSteeringBehavior steer = new CompoundSteeringBehavior(myAgent);
        BalancedCompoundSteeringBehavior nestedSteer = new BalancedCompoundSteeringBehavior(myAgent);
        nestedSteer.addSteerBehavior(steerBehavior1);
    steer.addSteerBehavior(nestedSteer);
    steer.addSteerBehavior(steerBehavior2);
mainBehavior.addBehavior(steer);
myAgent.setMainBehavior(mainBehavior);

Prioritizing behaviors

You can assign priority layers: The steering controller first checks the higher layer to see if all the behaviors returns a value higher than minLengthToInvalidSteer, if so it uses that layer. Otherwise, it moves on to the second layer, and so on.

To assign priority layers add behaviors with the following function:

  addSteerBehavior (AbstractSteeringBehavior behavior, int priority, float minLengthToInvalidSteer)

To optimize the process speed add the behaviors with the lowest priority first.

The layer and the min length to consider the behavior invalid are 0 by default.

Setting up forces

If a behavior extends from the AbstractStrengthSteeringBehavior class, you can manage how the produced forces will work.

Use setupStrengthControl(float scalar) to increase/decrease the steer force produced by a behavior or setupStrengthControl(Plane plane) If you want to work with 2D behaviors.

Example:

    Plane horizontalPlane = new Plane(new Vector3f(0,1,0), 0);

    steerBehavior1.setupStrengthControl(0.5f); //Force reduced a 50%
    steerBehavior2.setupStrengthControl(horizontalPlane); //Force contained in the XZ plane
    steerContainer.setupStrengthControl(horizontalPlane, 2f); //Contained in the XZ plane and increased a 100%

Implementing your own steer behavior

To benefit from all the features, you have to create a new class that extends from AbstractStrengthSteeringBehavior.

The responsible for the agent’s acceleration is the vector returned in the calculateRawSteering() method:

    @Override
    protected Vector3f calculateRawSteering() {
        Vector3f steerForce = Vector3f.ZERO;

        //calculations

        return steerForce;
    }

In addition, you can change a brake factor which will reduce the resultant velocity for the agent:

    @Override
    protected Vector3f calculateRawSteering(){
        this.setBrakingFactor(0.5f); //The agent's velocity will be reduced a 50%
        return Vector3f.ZERO;
    }

The braking force must be a float contained in the [0,1] interval

0 means the maximum braking force and 1 No braking force

Strict arguments

To ensure that the behavior will work as you had planned it to work It’s recommended to create your own IllegalArgumentException class. To do this, create your own container class extending from com.jme3.ai.agents.behaviors.npc.steering.SteeringExceptions; Each exception inside the container class extends from SteeringBehaviorException. Furthermore, It will help users to recognize better which is the origin of any problem.

Example:

    public class CustomSteeringExceptions extends SteeringExceptions  {

        public static class CustomRuntimeException extends SteeringBehaviorException {
            public CustomRuntimeException(String msg) { super(msg); }
        }

        // ... other exceptions ...
    }
    public SteerBehaviorConstructor(Agent agent, int value, Spatial spatial) {
        super(agent, spatial);
        if(value > 5) throw new CustomSteeringExceptions.customRuntimeException ("Value must be lower than 5");
        this.value = value;
    }

java steer behaviors project: jmesteer.bdevel.org