Game Theoretic Analysis of Road User Safety Scenarios Involving - - PowerPoint PPT Presentation

Game Theoretic Analysis of Road User Safety Scenarios Involving Autonomous Vehicles

Department of Information Engineering 11/09/2018 Umberto Michieli Leonardo Badia

Ri Rise of Au Autonomo mous Vehicles (AVs)

1

Ri Rise of Au Autonomo mous Vehicles (AVs)

1

Tr Transition to AVs

Smooth transition

2

Need to overcome many conflicts

PR PRO

• 1. Accidents ↓
• 2. Less stressful time
• 3. Less road congestion
• 4. Decreased emissions
• 5. Eventually faster than human-drivers

3

CO CONS

4

• 1. Social acceptance

CO CONS

• road regulations
• AVs are cautious
• 1. Social acceptance
• 2. Technological issues

RI RISK SK AVERSE RSE

• 3. Interactions w/ humans
• snow/rain
• yellow-lights
• partial occlusion

MO MOORE’S ’S LAW

4

Exp Expect cted Con Concl clusion

• ns:
• Game Theory extensions
• Accidents ↓ as share of AVs ↑
• Need for new traffic regulations
• Need for communication systems

Our Our Appr pproach

5

MODEL SIMULATION VALIDATION

Mo Modeling

Human-AVs interactions

Ga Game Th Theory (our approach)

Statistics

• Players have different utilities
• Distinguishable set of actions
• Statistical generality

Pr Propose sed mo models: s: 1. Cyclist vs. Vehicle on Zebra Crossing

• 2. Pedestrian vs. Vehicle

6

1.

• 1. Cy

Cyclist t vs. . Vehicle on Zebra Cr Crossing

Human Driver Cyclist Human Driver

8 15 6 1 15 7

• 400
• 400
• 500
• 200

20 7 Yield Walk Cycle

AV Cyclist AV

5 7 3 10 20 15

• 500
• 300

15 15

• 400
• 500

Nature

Go Go Go Stop Stop Stop

probability p

Yield Walk Cycle Go Go Go Stop Stop Stop

probability (1-p)

à SIMULTANEOUS BAYESIAN GAME COMMON KNOWLEDGE & FULL RATIONALITY

AI AIMS MS:

Cyclist vs. AV or human driver Accident rate curve as AVs ↑

7

TW TWO O PU PURE NEs

• 1. (CY, SG)
• 2. (CC, SS)

Human Driver Cyclist Human Driver

8 15 6 1 15 7

• 400
• 400
• 500
• 200

20 7 Yield Walk Cycle

AV Cyclist AV

5 7 3 10 20 15

• 500
• 300

15 15

• 400
• 500

Nature

Go Go Go Stop Stop Stop

probability p

Yield Walk Cycle Go Go Go Stop Stop Stop

probability (1-p)

ON ONE MIXED NE If If A AV: (C,S) If If h human: yield (p1) or cycle (1-p1), go (p2) or stop (1-p2) p1=93.7% p2=2.7%

1.

• 1. Cy

Cyclist t vs. . Vehicle on Zebra Cr Crossing

8

1.

• 1. Cy

Cyclist t vs. . Vehicle on Zebra Cr Crossing

% of Autonomous Vehicles % of Collisions

10 40 50 70 60 30 20 90 80 0.05 0.10 0.15 0.20 0.25 0.30 low speed medium speed high speed

9

% of Autonomous Vehicles % of Fatal Injuries

low speed medium speed high speed 10 20 30 40 50 60 70 80 90 0.02 0.04 0.06 0.08 0.10 0.12

1.

• 1. Cy

Cyclist v t vs. V . Vehicle o

• n Z

Zebra Cr Crossing

10

• 2. P

. Ped edes estri trian v

• vs. V

. Veh ehicle

Out Cross

Pedestrian Vehicle

Keep Brake (ta-tc’) (tc-ta) (tc-ta) (tc’-ta) (ta-tc)

PA PAYOF OFF IS TI TIME: ETA of

vehicle to make decision

11

1/ 1/ta 1/ 1/tc’ 1/ 1/tc moves at 1.4 .4 m/s la lane-width = 3.7 .75 m

ta < tc tc < ta < tc’ ta > tc’ CK CB O NE NE shif ifts:

32,23% 4,19% 63,58% 59,82% 21,07% 19,12% 0% 10% 20% 30% 40% 50% 60% 70% CROSS-KEEP CROSS-BRAKE OUT Human Driver AV

AV AVs hum human-dr drivers

• 2. P

. Ped edes estri trian v

• vs. V

. Veh ehicle

𝑤 ∼ max(𝒪(30,10), 0) km/h a = 2.5 m/s d ∼ 𝒱(10,50) m reaction time tr = 1.5 s à tc’ higher than AVs 𝑤 ∼ max(𝒪(50,10), 0) km/h }

AV AVs} huma uman n dri drivers rs SIMULATION PARAMETERS:

12

Co Conclusions à Fu Futu ture W e Work rk

13

Co Conclusions à Fu Futu ture W e Work rk

§ Game theory useful for human-AV interactions à improve realism

13

Co Conclusions à Fu Futu ture W e Work rk

§ Game theory useful for human-AV interactions à improve realism § Models are lightweight à embedding into communication systems and

13

traffic simulators

Co Conclusions à Fu Futu ture W e Work rk

§ Game theory useful for human-AV interactions à improve realism § Models are lightweight à embedding into communication systems and

13

traffic simulators § Accident rate ↓, dominance of pedestrians à new regulations needed, then new game analysis

Thank you for the attention!

Questions?