File size: 5,733 Bytes
6886206
7657290
 
 
6886206
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
7657290
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
6886206
92107a5
b960e8c
 
6886206
92107a5
6886206
92107a5
6886206
92107a5
b960e8c
92107a5
6886206
b960e8c
6886206
92107a5
6886206
92107a5
6886206
92107a5
6886206
 
92107a5
 
6886206
92107a5
 
 
7657290
 
92107a5
 
7657290
92107a5
 
 
7657290
 
92107a5
 
7657290
6886206
 
7657290
 
 
 
92107a5
7657290
 
 
 
6886206
 
7657290
 
6886206
7657290
 
 
 
 
6886206
7657290
 
 
 
 
 
6886206
 
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
from dataclasses import dataclass
from typing import Any

from openenv.core.rubrics import Rubric, WeightedSum

from .sim.world import World


@dataclass
class ScoreBreakdown:
    throughput: float
    emergency: float
    fairness: float
    efficiency: float
    planning: float
    safety: float
    total: float

    def as_dict(self) -> dict:
        return {
            "throughput": round(self.throughput, 4),
            "emergency": round(self.emergency, 4),
            "fairness": round(self.fairness, 4),
            "efficiency": round(self.efficiency, 4),
            "planning": round(self.planning, 4),
            "safety": round(self.safety, 4),
            "total": round(self.total, 4),
        }


class WorldRubric(Rubric):
    def forward(self, action: Any, observation: Any) -> float:
        world = observation
        return self.score(world)

    def score(self, world: World) -> float:
        raise NotImplementedError


class ThroughputRubric(WorldRubric):
    def score(self, world: World) -> float:
        spawned = world.metrics.spawned_civilian + world.metrics.spawned_emergency
        if spawned == 0:
            return 0.0
        cleared_vehicles = [v for v in world.vehicles.values() if v.cleared]
        frac = len(cleared_vehicles) / spawned
        if not cleared_vehicles:
            return 0.0
        slowdowns: list[float] = []
        for v in cleared_vehicles:
            optimal = sum(world.roads[rid].length for rid in v.route)
            actual = (v.clear_tick or world.tick) - v.spawn_tick
            slowdowns.append(actual / max(1, optimal))
        mean_slowdown = sum(slowdowns) / len(slowdowns)
        speed_score = max(0.0, min(1.0, 1.0 - (mean_slowdown - 1.0) / 0.8))
        return 0.3 * frac + 0.7 * speed_score


class EmergencyRubric(WorldRubric):
    def score(self, world: World) -> float:
        spawned = world.metrics.spawned_emergency
        if spawned == 0:
            return 1.0
        cleared = world.metrics.cleared_emergency
        clear_frac = cleared / spawned
        times = world.metrics.emergency_clear_times
        if not times:
            return 0.0
        mean_clear_ticks = sum(times) / len(times)
        budget_per_em = 40.0
        speed_score = max(0.0, 1.0 - mean_clear_ticks / budget_per_em)
        return 0.5 * clear_frac + 0.5 * (clear_frac * speed_score)


class FairnessRubric(WorldRubric):
    def score(self, world: World) -> float:
        budget = 150
        max_wait = world.metrics.max_wait_ticks_seen
        if max_wait <= 0:
            return 1.0
        return max(0.0, 1.0 - max_wait / budget)


class EfficiencyRubric(WorldRubric):
    def score(self, world: World) -> float:
        total_ticks_seen = max(1, world.tick * max(1, len(world.intersections)))
        wasted = world.metrics.wasted_green_ticks
        ratio = wasted / total_ticks_seen
        return max(0.0, 1.0 - 6.0 * ratio)


class PlanningRubric(WorldRubric):
    def score(self, world: World) -> float:
        budget = world.interventions_budget
        used = world.interventions_used
        invalid = world.metrics.invalid_actions
        if budget == 0:
            base = 1.0
        else:
            over = max(0, used - budget)
            base = 1.0 - (over / max(1, budget))
        penalty = min(1.0, invalid * 0.1)
        return max(0.0, base - penalty)


class SafetyRubric(WorldRubric):
    def score(self, world: World) -> float:
        if world.metrics.gridlock_events == 0:
            return 1.0
        return max(0.0, 1.0 - 0.5 * world.metrics.gridlock_events)


RUBRIC_CLASSES = {
    "throughput": ThroughputRubric,
    "emergency": EmergencyRubric,
    "fairness": FairnessRubric,
    "efficiency": EfficiencyRubric,
    "planning": PlanningRubric,
    "safety": SafetyRubric,
}

TASK_WEIGHTS: dict[str, dict[str, float]] = {
    "grid_balanced": {
        "throughput": 0.40,
        "emergency": 0.15,
        "fairness": 0.15,
        "efficiency": 0.15,
        "planning": 0.05,
        "safety": 0.10,
    },
    "demand_shift": {
        "throughput": 0.35,
        "emergency": 0.10,
        "fairness": 0.20,
        "efficiency": 0.15,
        "planning": 0.10,
        "safety": 0.10,
    },
    "incident_corridor": {
        "throughput": 0.15,
        "emergency": 0.40,
        "fairness": 0.10,
        "efficiency": 0.10,
        "planning": 0.15,
        "safety": 0.10,
    },
    "rush_hour_wave": {
        "throughput": 0.35,
        "emergency": 0.10,
        "fairness": 0.25,
        "efficiency": 0.10,
        "planning": 0.10,
        "safety": 0.10,
    },
    "multi_crisis": {
        "throughput": 0.15,
        "emergency": 0.30,
        "fairness": 0.10,
        "efficiency": 0.10,
        "planning": 0.20,
        "safety": 0.15,
    },
}

DIMENSION_ORDER = ["throughput", "emergency", "fairness", "efficiency", "planning", "safety"]


def build_rubric(task: str) -> WeightedSum:
    weights = TASK_WEIGHTS.get(task, TASK_WEIGHTS["grid_balanced"])
    rubrics = [RUBRIC_CLASSES[dim]() for dim in DIMENSION_ORDER]
    weight_list = [weights[dim] for dim in DIMENSION_ORDER]
    return WeightedSum(rubrics, weight_list)


def grade(world: World) -> ScoreBreakdown:
    rubric = build_rubric(world.task)
    total = rubric(None, world)
    total = max(0.0, min(1.0, total))

    scores = {}
    for i, dim in enumerate(DIMENSION_ORDER):
        scores[dim] = rubric._rubric_list[i].last_score

    return ScoreBreakdown(
        throughput=scores["throughput"],
        emergency=scores["emergency"],
        fairness=scores["fairness"],
        efficiency=scores["efficiency"],
        planning=scores["planning"],
        safety=scores["safety"],
        total=total,
    )