Susceptibility of Autonomous Driving Agents to Learning-Based Action-Space Attacks

Keyphrases

• Learning-based 100%
• Action Space 100%
• Autonomous Driving Agent 100%
• Processing pipeline 66%
• Deep Reinforcement Learning (deep RL) 66%
• Modular Processing 66%
• Behavioral Differences 33%
• Problem-based 33%
• Actuation 33%
• Resilience 33%
• Enhancement Mechanism 33%
• Valuable Insight 33%
• Making Decisions 33%
• Increasing Complexity 33%
• Inertial Measurement Unit 33%
• Safety-critical 33%
• Vehicle Safety 33%
• Assailant 33%
• Intelligent Vehicle 33%
• Camera Measurements 33%
• Vehicle State 33%
• Cybersecurity Risk 33%
• Reinforcement Learning Problems 33%
• Adversarial Training 33%
• Driving Agent 33%
• Autonomous Driving System 33%
• Side Collision 33%
• Progressive Neural Networks 33%
• Direct Risk 33%
• End-to-end Driving Model 33%

Engineering

• Action Space 100%
• Reinforcement Learning 66%
• Actuation 33%
• Limitations 33%
• Future Design 33%
• Inertial Measurement 33%
• Units of Measurement 33%
• Vehicle Safety 33%
• Intelligent Vehicle Highway Systems 33%
• Critical Moment 33%

Computer Science

• Autonomous Driving 100%
• Deep Reinforcement Learning 66%
• Neural Network 33%
• Learning Problem 33%
• Cybersecurity 33%
• Measurement Unit 33%
• Adversarial Machine Learning 33%