Advanced Computer Graphics

Transcription

1 Advanced Computer Graphics Lecture 14: Game AI Techniques Bernhard Jung TU-BAF, Summer 2007 Overview Components of Game AI Systems Animation Movement & Pathfinding Behaviors Decision Making Finite State Machines Decision Trees Rule-based Systems Literature Lars Lidén: Strategic and Tactical Reasoning with Waypoints. In Steve Rabin (Ed): AI Programming Wisdom, Charles River Media, pp , Alex J. Champandard. Synthetic Creatures with Learning and Reactive Behaviors. New Riders Games. Amit s Game Programming Information. Presentation based on Lars Liden. The Use of Artificial Intelligence in the Computer Game Industry 1

2 Introduction The term A.I. is used broadly in the computer game industry Games come in many genres and the techniques employed by the different genres vary widely Will be discussing A.I. for non-player characters (NPCs) in a firstperson or third-person shooter Fully rendered 3D environment Player controls a single character Won t be discussing use of AI for: Board games (Chess, Go) Strategy games (Civilization / Age of Empires) Camera control in 3D games Simulation games (The Sims, Sim City) Racing games Components of an AI System Animation Responsible for controlling NPC body Movement Responsible for controlling NPC navigation Behavior Responsible for controlling NPC decision making 2

3 AI Components: Animation NPC models built by artists Use tools such as 3D Studio Max or Maya Models are are constructed from bones Bones are connected by articulated joints The skeletal system is covered by a mesh of textured polygons AI Components: Animation Animation sequences are generated by defining how joints should articulate through time Animation sequences for a model are either: Hand generated by a computer animator Recorded from real human (or animal) movements and applied to a skeletal system ( motion capture ) Walking sequence: 3

4 AI Components: Animation Animation sequences tend to be: Motion primitives: Run, Walk, Jump, Side-step, Climb Transitions Start_Walk, Run_To_Jump, Jump_Land Some animation sequences only take control of part of the body: wave_hello hand_signal_stop swing_ice_axe AI Components: Animation Animation Selection Parameterize Animation Selected Joint Control Inverse Kinematics 4

5 AI Components: Animation First step in A.I. is to select which animation sequence or sequences should be applied to a model Many influences: Desired behavior chosen by decision system What animation is currently playing The current velocity and direction of the NPC The terrain the NPC is standing on AI Components: Animation Second step is to parameterize animations Speed up or slow down animation Slow walk, fast walk Accelerate / decelerate stop and start of run Slow run as approach sharp turn Blend between animations walk-to-run 70% normal walk + 30% limp Layer animations Mix hand_wave on top of walk animation 5

6 AI Components: Animation Next might add selected Joint Control Take control of particular joints Either: Ignore joint motion in pre-generated animation Blend with pre-generated joint motion Used for: Head Turning Looking at a particular object or location Arm aiming Point gun at a location AI Components: Animation And finally, add inverse kinematics Algorithmically determine the joint configuration required for an end-effector (hand or foot) to reach a particular location Used for: Keep the feet on the ground on uneven terrain or when walking up stairs Reaching hand out to open a door, pick up and object 6

7 Movement: Pathfinding The primary component of a movement system is pathfinding. Find a route for an NPC to travel from one location in the world to another. Route depends on: NPC s size Will NPC s body fit in a given location? NPC s navigation ability Walk, Jump, Climb, Swim Movement: Pathfinding Tools Waypoint Position in a map that is used for navigation Usually placed in world manually by a level designer Link Connection between two waypoints Often annotated with the required navigation type (Jump, Swim, Climb) For a given NPC, two waypoints are linked when: The NPC has room enough to move from one node to another without colliding with the world geometry The NPC has the required navigation ability Node Graph Data structure holding all waypoints and links Either generated manually by a level designer or automatically by the computer and annotated by a level designer 7

8 Movement: Node Graph Movement: Route Generation A* is the preferred pathfinding algorithm for quickly finding a short path between two waypoints heuristic search algorithm incorporates estimate of distance to goal guaranteed to find optimal path if link costs are never overestimated Often modified for other route types: Fastest Route Shortest route is not often the fastest as it may take an NPC longer to climb a ladder or swim between two waypoints than to run or walk Safest Route For tactical reasons an NPC may choose to prefer a longer less exposed route than a shorter or faster route with no cover see: applet: ; lokale Kopie 8

9 Movement: Obstacle Avoidance Often the character must avoid players, other characters, and vehicles that are moving rapidly through the environment. Characters must not get stuck on each other at choke points, and they must maintain enough spacing to maneuver when traveling in groups. Obstacle avoidance may be implemented using trajectory prediction and layered steering behaviors. AI Components: Behavior Responsible for deciding what behavior an NPC should express Sense Each NPC has a think cycle: Sense Environment Select Behavior Express behavior Think Act 9

10 AI Components: Behavior Sense Environment What to I see? What can I hear / smell? Includes internal states: How tired am I? What s my health? Select Behavior Finite State Machine Decision Tree Rule-based System Express Behavior Select appropriate animation sequences Quake Shambler Monster FSM 10

11 Quake Dog Monster Hierarchical FSM Quake Decision Tree Just one tree Actions: Attack, Retreat, Chase, Spawn, Wander Could add additional trees: If I m attacking, which weapon should I use? If I m wandering, which way should I go? Much like hierarchical FSMs 11

12 Production (Rule-based) Systems if-then rules if C1, C2, then A1, A2, "deductive" rule-based systems: if- and then-parts are assertions (facts); thenparts are added to working memory when the rule is fired rule-based reaction systems: thenparts are arbitrary actions; e.g. assert a fact in working memory, execute a procedure Working Memory (contains facts) P Q Rule base (contains rules) R1 R2 P Q P Psychological plausibility e.g. Newell & Simon, 1972 "cognitive architectures", e.g. SOAR, ACT-R Rules are "modular" easy modification of single rules possible Rule Interpreter Decision Making Rule Based Systems TASKS: Simple things for a NPC to do, such as: Turn to face a location or an object (TASK_FACE_FRIEND) Find a path to a location (TASK_FIND_PATH_TO_ENEMY) Move along a path (TASK_WALK_PATH) Task-1 Schedule Task-2 Task-3 SCHEDULES: Named lists of tasks Selected by logic system Performed by NPC until: All tasks have finished One task has failed Schedule has been Interrupted SCHEDULE_GET_WEAPON TASK_FIND_WEAPON TASK_FIND_PATH TASK_RUN_PATH TASK_PICKUP_WEAPON SCHEDULE_FLANK_ATTACK TASK_FIND_FLANK_POSITION TASK_FIND_PATH TASK_RUN_PATH TASK_RANGE_ATTACK 12

13 Decision Making: Rule-Based Systems Conditions Predicates that are set every time an NPC thinks Conditions serve two purposes: Schedule Selection Schedule Interruption Schedule selection is rule based: If (CONDITION_HEAR_DANGER) and not (CONDITION_HAVE_WEAPON) then select schedule (SCHEDULE_GET_WEAPON) If (CONDITION_HAVE_WEAPON) and (CONDITION_OUT_OF_AMMO) then select schedule (SCHEDULE_RELOAD_WEAPON) Decision Making: Rule-Based Systems Interrupt conditions Schedules also contain interrupt conditions Same set of predicates used for schedule selection If interrupt condition is listed in the schedule, and the condition is true, the schedule will be interrupted Example SCHEDULE_GET_WEAPON TASK_FIND_WEAPON TASK_FIND_PATH TASK_RUN_PATH TASK_PICKUP_WEAPON CONDITION_HEAVY_DAMAGE CONDITION_ENEMY_DEAD 13

14 Decision Making: Rule-Based Systems Task Failure Sometimes a task can t be completed For example: (TASK_FIND_WEAPON) when no weapon available (TASK_FIND_PATH) when no way to get to destination In such cases the schedule will exit pre-maturely Fail Schedules: Can specify a fail schedule e.g. take cover schedule chosen when any of the tasks fail SCHEDULE_GET_WEAPON TASK_SET_FAIL_SCHEDULE SCHEDULE_TAKE_COVER TASK_FIND_WEAPON TASK_FIND_PATH TASK_RUN_PATH TASK_PICKUP_WEAPON Decision Making: Rule-Based Systems NPC Think Cycle: Update predicate values (conditions) If any conditions interrupt the current schedule, select a new schedule Perform next task in schedule list If task fails, go to fail schedule (if defined) or select a new schedule If all tasks have been completed, select a new schedule 14

15 Decision Making: Rule-Based System Another Example Rules model the behavior of a dog interacting with a human. The rules are contextualized, i.e. apply only in certain states ("goals") Mike Christian: A Simple Inference Engine for a Rule-Based Architecture. In AI Game Programming Wisdom. Charles River Media BRAIN_START(Dog, NPCBrainLib) GOAL (Nap) IF (HearsNoise) GOTO (Investigate) GOAL (Investigate) IF (SeesHuman) GOTO (ConsiderAction) IF (Done) GOTO (Nap) GOAL (ConsiderAction) IF (LikesHuman) GOTO (Follow) IF (WaryOfHuman) GOTO (RunAway) IF (HateHuman) GOTO (Attack) GOAL (Follow) IF (Close) GOSUB (Beg) IF (Tired) GOSUB (Rest) IF (HumanGone) GOTO (GoHome) GOAL (Rest) IF (Done) RETURN GOAL (Beg) IF (GetFood) GOSUB (Eat) GOAL (Eat) GOAL (GoHome) BRAIN_END Decision Making: Rule-Based System Yet Another Example Rule-based System for NPC capabilities on several levels: Low-level movement behaviors such as obstacle avoidance, chasing and evading, Weapon selection based on the current spatial position and situation of the game, Attacking the enemy using prediction of movement, basic aiming and firing, Tactical reasoning and selection of the right action in the current context. 15

16 Decision Making: Rule-Based System Yet Another Example Senses (represented by symbols) Range sensors Left Obstacle, Front Obstacle, Right Obstacle Range sensors are used to determine the distance of the closest obstacles. The default parameters check the space within three steps of the animat's position, returning true if the area is clear. The sensors are placed at -45, 0 and +45. Collision Collision Presence of enemy Enemy, Close, Far Presence of Health & Armor Items Health Item, Armor Item Personal state Low Health, Full Health, Low Armor, Full Armor Decision Making: Rule-Based System Yet Another Example Actions Movement Forward Move in the direction the body is facing. Seek Head towards the enemy. Flee Move away from the enemy. Side Take a lateral step as a strafing motion. Avoid Moves away from the obstacle that was hit, in the direction of the normal of the collision.* View Control Look Right, Look Left Rotate the view by +/-30. Look Behind Turn the view sharply by over 90 left or right. Look to Enemy/Health/Armor Turn to face this entity in the field of view. Look Around Gaze around by rotating the view parametrically.* Weapon Control Use Weapon Four commands to select the most popular weapons: rocket launcher, railgun, chaingun and hyperblaster. Fire Pull the trigger on the current weapon. * avoid and look around behaviors cannot be implemented with a symbolic rulebased system; actions commands trigger respective behavior procedures 16

17 Decision Making: Rule-Based System Yet Another Example Production Rules Action Selection IF enemy AND health_low THEN retreat IF enemy AND NOT distance_close THEN pursue IF enemy THEN dodge IF true THEN NOT retreat AND NOT pursue AND NOT dodge Movement IF collision AND enemy THEN move_avoid IF retreat THEN move_flee IF pursue THEN move_seek IF dodge THEN move_side IF true THEN move_forward Weapon Selection IF enemy AND distance_far THEN use_rocketlauncher AND use_railgun IF enemy AND NOT distance_far THEN use_chaingun AND use_hyperblaster View Control IF enemy THEN look_enemy AND fire IF obstacle_front AND obstacle_left AND obstacle_right THEN look_behind IF obstacle_front AND obstacle_left THEN look_right IF obstacle_front AND obstacle_right THEN look_left IF NOT health_full AND health_item THEN look_health IF NOT armor_full AND armor_item THEN look_armor IF true THEN look around Decision Making: SOAR Quake Bot Developing an Artificial Intelligence Engine. van Lent; Laird (1999) Quake bot decision cycle Perceive: Accept sensor information from the game Think: Select and execute relevant knowledge Act: Execute actions in the game. 17

18 Decision Making: SOAR Quake Bot Developing an Artificial Intelligence Engine. van Lent; Laird (1999) QuakeBot Sensor information & actions Decision Making Rule-based System: Good and Bad Advantages Corresponds to way people often think of knowledge Very expressive and allows very complex behaviors Modular knowledge Easy to write and debug compared to decision trees More concise than FSM Disadvantages Can be memory intensive Can be computationally intensive Sometimes difficult to debug 18