Robot Rumble ALPHA
discord
try it!
boards
tutorial
docs
login
/
signup
# # Unsupported browser type! # The Garage officially supports Chrome 85+, Microsoft Edge 85+, Firefox 78+, Opera 71+ # The Garage DOES NOT support Safari # Your browser is: Safari 16.3 # # If you cannot switch to a different browser, consider downloading Rumblebot, our command line tool # https://rr-docs.readthedocs.io/en/latest/rumblebot.html # # Unsupported browser type! # The Garage officially supports Chrome 85+, Microsoft Edge 85+, Firefox 78+, Opera 71+ # The Garage DOES NOT support Safari # Your browser is: Safari 16.3 # # If you cannot switch to a different browser, consider downloading Rumblebot, our command line tool # https://rr-docs.readthedocs.io/en/latest/rumblebot.html # from rumblelib import * from typing import ( Dict, NamedTuple, Sequence, Hashable, List, Optional, Tuple, Set, ) from collections import namedtuple import random ############################################################ ### To implement new strategies rewrite make_obj_plan()! ### ############################################################ robot_actions: Dict[ str, Action ] = {} # Global variable to store all actions in init_turn() and to be used in robot() shift_to_direction: Dict[Tuple, Direction] = { (1, 0): Direction.East, (-1, 0): Direction.West, (0, 1): Direction.South, (0, -1): Direction.North, } # A plan contains an .action for an .id # Plans are selected by *lowest* .score such that for each .target there is only one selection Plan = namedtuple( "Plan", [ "id", # str: id of object "target", # Hashable: only one target will be selected from all equal targets "score", # float: minimum score is highest priority "action", # Action: action to be executed]) ], ) def make_info(state: State) -> Dict: """ Return anything you want to pre-calculate at the beginning of a turn """ return {} def make_obj_plan(state: State, info: Dict, obj: Obj) -> Sequence[Plan]: """ Should emit a list of plans for the current object """ result: List[Plan] = [] enemies = state.objs_by_team(state.other_team) if not enemies: return [] closest_enemies, distance = find_closest(obj, enemies, get_dist) closest_enemy = random.choice(closest_enemies) shifts = get_directions_to(closest_enemy.coords - obj.coords) for shift in shifts: assert shift != (0, 0) # should never happen next_pos = obj.coords + Coords(shift.x, shift.y) if is_spawn_turn(state.turn + 1) and not is_inside_nonspawn(next_pos): continue direction = shift_to_direction[shift] next_obj = state.obj_by_coords(next_pos) if next_obj is not None and next_obj.team == state.other_team: action = Action.attack(direction) else: action = Action.move(direction) plan = Plan( id=obj.id, target=next_pos, score=distance, action=action, ) result.append(plan) return result ############################################################## ### Following functions probably do not have to be changed ### ############################################################## Shift = namedtuple("Shift", "x y") def init_turn(state: State): """ run once for each turn use this to initialize global variables to be used by all units """ global robot_actions try: info = make_info(state) plans = make_plans(state, info) selected_actions = select_plans(plans) robot_actions = selected_actions except Exception as exc: import traceback traceback.print_exc() def robot(state: State, unit: Obj) -> Optional[Action]: """ called for each bot and needs to return an action """ return robot_actions.get(unit.id) # bots without plans will return None and be idle def make_plans(state: State, info: Dict) -> List[Plan]: plans: List[Plan] = [] for obj in state.objs_by_team(state.our_team): plans.extend(make_obj_plan(state, info, obj)) return plans def select_plans(plans: List[Plan]) -> Dict[str, Action]: # Simple greedy selection by lowest score targets_used: Set[Hashable] = set() random.shuffle(plans) # to avoid systematic priority effects plans = sorted(plans, key=lambda x: x.score) result: Dict[str, Action] = {} for plan in plans: if plan.id in result or ( plan.target is not None and plan.target in targets_used ): continue result[plan.id] = plan.action targets_used.add(plan.target) return result def dist_from_center(coord: Coords): # "octagonal distance" dx = coord.x - 9 dy = coord.y - 9 return max(abs(dx), abs(dy), abs(dx) + abs(dy) - 4) def is_spawn_turn(turn: int) -> bool: return turn % 10 == 0 def is_inside_field(coord: Coords): return dist_from_center(coord) <= 8 def is_inside_nonspawn(coord: Coords): return dist_from_center(coord) < 8 def is_spawn_region(coord: Coords): return dist_from_center(coord) == 8 def get_dist(obj1: Obj, obj2: Obj) -> int: return obj1.coords.walking_distance_to(obj2.coords) def find_closest_idx(obj, others, dist_func) -> Tuple[List[int], Optional[int]]: """ Return indices of position from poses2 which is closest to pos1 Also return distance """ if not others: return [], None distances = [dist_func(obj, other) for other in others] shortest_distance = min(distances) indices = [ i for i, distance in enumerate(distances) if distance == shortest_distance ] return ( indices, shortest_distance, ) def find_closest(obj, others, dist_func): closest_indices, distance = find_closest_idx(obj, others, get_dist) return [others[i] for i in closest_indices], distance def get_directions_to(shift: Shift) -> Set[Shift]: """ Returns all directions that would bring you closer along shift """ if shift == (0, 0): return {Shift(0, 0)} result: Set[Shift] = set() if shift.x > 0: result.add(Shift(1, 0)) if shift.x < 0: result.add(Shift(-1, 0)) if shift.y > 0: result.add(Shift(0, 1)) if shift.y < 0: result.add(Shift(0, -1)) return result
Made with <3 by Anton and Noa
github org