Waymo Combines LiDAR and Radar for Self-Driving Safety Amid AI Exploration

Waymo, Alphabet Inc.’s self-driving technology division, is exploring the use of generative artificial intelligence (AI) as part of its broader efforts to enhance autonomous driving capabilities. However, a senior executive at the company, Srikanth Thirumalai, emphasized that traditional sensor technologies such as LiDAR and radar remain central to ensuring safety in all driving conditions. Thirumalai made the remarks during an interview at the AI4 Conference in Las Vegas, where he discussed the ongoing development of Waymo’s autonomous systems [1].

According to Thirumalai, the company has conducted extensive research into what works at scale and has found that the combination of LiDAR and radar provides a critical safety net that is particularly valuable in unpredictable environments and extreme weather. This approach contrasts with the camera-only strategy employed by some competitors, such as Tesla. During a presentation at the conference, Thirumalai demonstrated how Waymo’s sensors detected pedestrian movements and potential hazards in scenarios where cameras failed to capture the same level of detail. These examples underscore the company’s commitment to maintaining a robust, multi-sensor approach [1].

Waymo has also been experimenting with generative AI models, including Google’s Gemini, as part of its broader AI stack. The company has explored the use of multimodal models such as EMMA, an end-to-end model designed for autonomous driving. While co-training with these models has shown potential in areas such as object detection and road mapping, Thirumalai noted that EMMA is still limited in its ability to process large volumes of data and does not incorporate LiDAR or radar. These limitations, he said, present significant challenges to using generative AI as a standalone system for driving [1].

Despite these challenges, Waymo continues to invest in AI research, recognizing its potential to improve perception and decision-making in autonomous systems. Thirumalai described the integration of generative AI as an area of “intense research,” but stressed that much work remains to simplify and optimize the system for real-world applications. The company’s approach reflects a broader industry trend toward balancing innovation with reliability, particularly as self-driving systems are deployed in increasingly complex environments [1].

The contrast between Waymo’s sensor-heavy approach and the camera-based strategy of Tesla highlights a key debate in the self-driving industry. While Tesla’s Elon Musk has long dismissed LiDAR as unnecessary, Waymo’s leadership argues that the technology remains essential for achieving the highest level of safety. Thirumalai declined to directly compare the safety of different self-driving systems, noting that the lack of standardized metrics makes such comparisons difficult. However, he pointed to Waymo’s 100 million driverless miles as a demonstration of its safety at scale [1].

Waymo’s emphasis on sensor fusion and cautious integration of AI reflects a methodical approach to scaling autonomous driving. As the company expands its operations to five cities by mid-2025, the balance between cutting-edge AI and proven sensor technologies will likely remain a defining factor in its success.

Source: [1] Waymo engineering exec says LiDAR and radar sensors ... (https://fortune.com/2025/08/15/waymo-srikanth-thirumalai-interview-ai4-conference-las-vegas-lidar-radar-self-driving-safety-tesla/)