File size: 11,108 Bytes
051c46f | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 | """Next Passage: time-constrained grid navigation with periodic doors and moving obstacles.
Simulation only. One cell per step (4-neighbour moves or wait).
* Doors open and close on a known periodic schedule.
* Obstacles patrol back and forth along corridors and stall at random (probability p_stall per step),
so their future positions are only approximately predictable. They do not avoid the robot.
Planners
* static : BFS shortest path on the static map (doors treated as open); waits if the next cell is blocked now.
* reactive : like static, but re-runs BFS around currently blocked cells whenever it is blocked.
* timed : Next Passage. Space-time A* over (cell, time) with wait actions, using the door schedule and
predicted obstacle motion (cells occupied at t and t-1 are avoided). Replans when an observed
obstacle deviates from prediction or the next step became invalid.
A line-by-line JavaScript port lives in web/core.js; parity is tested on fixed seeds.
"""
from nlrt import Mulberry32
W, H = 21, 15
MOVES = [(0, 0), (1, 0), (-1, 0), (0, 1), (0, -1)]
DENSITY = {"low": 3, "medium": 6, "high": 10}
class Heap:
"""Binary min-heap on tuples; identical algorithm in JS so pop order matches exactly."""
def __init__(self):
self.a = []
def push(self, item):
a = self.a
a.append(item)
i = len(a) - 1
while i > 0:
p = (i - 1) // 2
if a[p] <= a[i]:
break
a[p], a[i] = a[i], a[p]
i = p
def pop(self):
a = self.a
top = a[0]
last = a.pop()
if a:
a[0] = last
i, n = 0, len(a)
while True:
l, r, m = 2 * i + 1, 2 * i + 2, i
if l < n and a[l] < a[m]:
m = l
if r < n and a[r] < a[m]:
m = r
if m == i:
break
a[m], a[i] = a[i], a[m]
i = m
return top
def __len__(self):
return len(self.a)
def make_world(seed, density="medium", p_stall=0.15):
rng = Mulberry32(seed * 2654435761 % 4294967296)
walls = set()
for x in range(W):
walls.add((x, 0)); walls.add((x, H - 1))
for y in range(H):
walls.add((0, y)); walls.add((W - 1, y))
doors = []
for wx in (5, 10, 15):
gap_open = 1 + rng.randint(H - 2)
gap_door = 1 + rng.randint(H - 2)
while abs(gap_door - gap_open) < 3:
gap_door = 1 + rng.randint(H - 2)
for y in range(1, H - 1):
if y not in (gap_open, gap_door):
walls.add((wx, y))
period = 8 + rng.randint(8)
doors.append({"cell": [wx, gap_door], "period": period, "phase": rng.randint(period),
"closed": max(2, period // 2 - 1 + rng.randint(3))})
for _ in range(6): # pillars
x, y = 2 + rng.randint(W - 4), 2 + rng.randint(H - 4)
if x not in (4, 5, 6, 9, 10, 11, 14, 15, 16):
walls.add((x, y))
start, goal = [1, 1 + rng.randint(H - 2)], [W - 2, 1 + rng.randint(H - 2)]
walls.discard(tuple(start)); walls.discard(tuple(goal))
door_cells = {tuple(d["cell"]) for d in doors}
obstacles, used = [], set()
tries = 0
while len(obstacles) < DENSITY[density] and tries < 400:
tries += 1
horiz = rng.random() < 0.5
x, y = 1 + rng.randint(W - 2), 1 + rng.randint(H - 2)
length = 4 + rng.randint(7)
cells = []
for k in range(length):
c = (x + k, y) if horiz else (x, y + k)
if c in walls or c in door_cells or c in used or c[0] >= W - 1 or c[1] >= H - 1 \
or list(c) == start or list(c) == goal:
break
cells.append(c)
if len(cells) < 3:
continue
loop = cells + cells[-2:0:-1]
used.update(cells)
obstacles.append({"loop": [list(c) for c in loop], "offset": rng.randint(len(loop))})
return {"seed": seed, "density": density, "p_stall": p_stall, "walls": sorted(list(c) for c in walls),
"doors": doors, "obstacles": obstacles, "start": start, "goal": goal}
def door_closed(d, t):
return (t + d["phase"]) % d["period"] < d["closed"]
def free(world_walls, c):
return 0 <= c[0] < W and 0 <= c[1] < H and c not in world_walls
def bfs(walls, start, goal, blocked=frozenset()):
"""Return (path, expanded). Neighbour order follows MOVES[1:]."""
start, goal = tuple(start), tuple(goal)
prev = {start: None}
q, qi, expanded = [start], 0, 0
while qi < len(q):
c = q[qi]; qi += 1; expanded += 1
if c == goal:
path = []
while c is not None:
path.append(c); c = prev[c]
return path[::-1], expanded
for dx, dy in MOVES[1:]:
n = (c[0] + dx, c[1] + dy)
if n not in prev and free(walls, n) and n not in blocked:
prev[n] = c
q.append(n)
return None, expanded
def distance_map(walls, goal):
goal = tuple(goal)
dist = {goal: 0}
q, qi = [goal], 0
while qi < len(q):
c = q[qi]; qi += 1
for dx, dy in MOVES[1:]:
n = (c[0] + dx, c[1] + dy)
if n not in dist and free(walls, n):
dist[n] = dist[c] + 1
q.append(n)
return dist
def timed_astar(walls, doors, dist, start, t0, goal, obs_idx, loops, horizon=160, max_expand=60000):
"""Space-time A*. obs_idx: current loop index of each obstacle (prediction assumes no stalls)."""
goal = tuple(goal)
door_at = {tuple(d["cell"]): d for d in doors}
def occupied(t):
cells = set()
for i, L in enumerate(loops):
for tt in (t, t - 1):
if tt >= t0:
cells.add(tuple(L[(obs_idx[i] + tt - t0) % len(L)]))
return cells
occ_cache = {}
start = tuple(start)
if start not in dist:
return None, 0
heap = Heap()
seq = 0
heap.push((dist[start], 0, seq, start[0], start[1], t0))
parent = {(start, t0): None}
expanded = 0
while len(heap) and expanded < max_expand:
f, negg, _, x, y, t = heap.pop()
c = (x, y)
expanded += 1
if c == goal:
path, node = [], (c, t)
while node is not None:
path.append(list(node[0])); node = parent[node]
return path[::-1], expanded
if t - t0 >= horizon:
continue
nt = t + 1
if nt not in occ_cache:
occ_cache[nt] = occupied(nt)
occ = occ_cache[nt]
for dx, dy in MOVES:
n = (x + dx, y + dy)
if not free(walls, n) or n in occ or n not in dist:
continue
d = door_at.get(n)
if d is not None and door_closed(d, nt):
continue
key = (n, nt)
if key in parent:
continue
parent[key] = (c, t)
seq += 1
g = nt - t0
heap.push((g + dist[n], -g, seq, n[0], n[1], nt))
return None, expanded
def run_episode(world, planner, max_steps=None):
walls = {tuple(c) for c in world["walls"]}
doors = world["doors"]
door_at = {tuple(d["cell"]): d for d in doors}
loops = [o["loop"] for o in world["obstacles"]]
idx = [o["offset"] % len(o["loop"]) for o in world["obstacles"]]
dist = distance_map(walls, world["goal"])
static_path, nodes = bfs(walls, world["start"], world["goal"])
max_steps = max_steps or 3 * len(static_path) + 60
rng = Mulberry32(world["seed"] * 7 + 3)
pos, t = tuple(world["start"]), 0
goal = tuple(world["goal"])
plan, plan_i, replans, collisions, door_hits, hard_waits = list(static_path), 0, 0, 0, 0, 0
pred = list(idx)
trace = [{"t": 0, "robot": list(pos), "obs": [loops[i][idx[i]] for i in range(len(loops))]}]
events = []
if planner == "timed":
plan, n = timed_astar(walls, doors, dist, pos, 0, goal, idx, loops)
nodes += n; replans += 1; plan_i = 0
arrived = None
while t < max_steps:
if pos == goal:
arrived = t
break
obs_now = {tuple(loops[i][idx[i]]) for i in range(len(loops))}
nxt = pos
if planner in ("static", "reactive"):
if plan is not None and plan_i + 1 < len(plan):
cand = tuple(plan[plan_i + 1])
d = door_at.get(cand)
blocked = cand in obs_now or (d is not None and door_closed(d, t))
if not blocked:
nxt = cand
plan_i += 1
elif planner == "reactive":
block = set(obs_now) | {tuple(dd["cell"]) for dd in doors if door_closed(dd, t)}
newp, n = bfs(walls, pos, goal, frozenset(block - {pos}))
nodes += n; replans += 1
if newp is not None and len(newp) > 1:
plan, plan_i = newp, 1
nxt = tuple(newp[1])
else:
hard_waits += 1
else:
hard_waits += 1
else:
deviated = any(idx[i] != pred[i] for i in range(len(loops)))
invalid = plan is None or plan_i + 1 >= len(plan)
if deviated or invalid:
plan, n = timed_astar(walls, doors, dist, pos, t, goal, idx, loops)
nodes += n; replans += 1; plan_i = 0
pred = list(idx)
if plan is not None and plan_i + 1 < len(plan):
nxt = tuple(plan[plan_i + 1]); plan_i += 1
else:
hard_waits += 1
# obstacles move (stall with probability p_stall), then robot moves
old = [tuple(loops[i][idx[i]]) for i in range(len(loops))]
for i in range(len(loops)):
if rng.random() >= world["p_stall"]:
idx[i] = (idx[i] + 1) % len(loops[i])
pred[i] = (pred[i] + 1) % len(loops[i])
t += 1
new = [tuple(loops[i][idx[i]]) for i in range(len(loops))]
hit = any(new[i] == nxt or (old[i] == nxt and new[i] == pos) for i in range(len(loops)))
prev_pos, pos = pos, nxt
if hit:
collisions += 1
events.append({"t": t, "type": "collision", "cell": list(pos)})
d = door_at.get(pos)
if d is not None and door_closed(d, t):
door_hits += 1
events.append({"t": t, "type": "door_closed_on_robot", "cell": list(pos)})
trace.append({"t": t, "robot": list(pos), "obs": [list(c) for c in new]})
if arrived is None and pos == goal:
arrived = t
return {"arrived": arrived is not None, "arrival_time": arrived, "collisions": collisions, "door_hits": door_hits,
"replans": replans, "nodes_expanded": nodes, "waits_blocked": hard_waits, "static_length": len(static_path) - 1,
"max_steps": max_steps, "events": events, "trace": trace}
|