A groundbreaking study led by the University of California, Irvine, reveals critical vulnerabilities in LiDAR (Light Detection and Ranging) technology, a cornerstone of many autonomous vehicles' navigation systems. The research, conducted in collaboration with Japan's Keio University, highlights the potential risks associated with spoofing attacks, which could lead to unsafe driving behaviors and collisions.
Presented at the Network and Distributed System Security Symposium, the study conducted by lead author Takami Sato, a UCI Ph.D. candidate in computer science, sheds light on the susceptibility of commercially available LiDAR systems to manipulation. By employing lasers, the researchers demonstrated the ability to deceive LiDAR into detecting nonexistent objects while overlooking genuine obstacles—a flaw that could result in unwarranted braking or collisions.
Sato emphasized the significance of their investigation, labeling it as the most extensive examination of LiDAR vulnerabilities to date. Through a combination of real-world testing and computer modeling, the team identified 15 new findings that underscore the need for enhanced safety measures in future autonomous vehicle systems.
LiDAR, a key technology utilized by leading companies such as Google's Waymo and General Motors's Cruise, is also integral to consumer-operated models from Volvo, Mercedes-Benz, and Huawei. Despite advancements in newer LiDAR iterations, the study uncovered potential weaknesses in both first-generation and subsequent versions.
The researchers demonstrated a "fake object injection" attack on first-generation LiDAR systems, tricking sensors into detecting nonexistent pedestrians or other vehicles, thereby triggering unsafe behaviors like emergency braking. Although newer LiDAR models employ countermeasures such as timing randomization, the study revealed novel methods to circumvent these defenses.
Using a custom laser apparatus, the team successfully concealed existing vehicles from next-generation LiDAR sensors, illustrating the vulnerability of these advanced systems to sophisticated spoofing attacks.
Qi Alfred Chen, senior co-author and UCI assistant professor of computer science, emphasized the gravity of the findings, warning that such attacks could directly induce hazardous driving behaviors in autonomous vehicles, including emergency braking and collisions.
The study underscores the urgent need for robust security measures to safeguard LiDAR technology and ensure the safety of autonomous vehicles on our roads. As autonomous driving technology continues to evolve, addressing these vulnerabilities is paramount to realizing the full potential of self-driving vehicles while prioritizing passenger safety.
More: https://techxplore.com/news/2024-03-autonomous-vehicle-technology-vulnerable-road.html
