As sensor bandwidths grow and autonomy demands real-time environmental understanding, the pressure on SoC design continues to intensify. The presentation explores how automotive-grade compute platforms are being architected to handle high-bandwidth, multimodal sensor fusion – balancing latency, scalability and safety. From CPU/GPU/NPU trade-offs to data movement bottlenecks, it will examine how Tier 1s and chipmakers are aligning compute architectures with the rigorous demands of L3+ systems.

• Trade-offs in compute architectures for sensor fusion
• Low-latency fusion of LiDAR, radar, camera, and ultrasonic data
• Scalable, safety-compliant platforms for real-time perception
• Optimized SoC design to overcome data movement bottlenecks