GTC 2020: Ultra-Fast Radar Simulation for Radar System Design and Automotive Applications
After clicking “Watch Now” you will be prompted to login or join.
Click “Watch Now” to login or join the NVIDIA Developer Program.
Ultra-Fast Radar Simulation for Radar System Design and Automotive Applications
Jeff Decker, ANSYS
Ultra-fast, physics-based radar simulations are required to design and deploy hardware-in-the-loop systems for applications such as autonomous vehicles and driver assistance systems (ADAS/AV). Rapid radar image generation is also an enabling technology for AI algorithms, vastly expanding the data sets to train and test the algorithms. We're developing an ultra-fast, end-to-end, GPU-accelerated radar image simulation engine for automotive, AI, and other applications. The shooting and bouncing rays (SBR) technique generates physics-based range-Doppler images of dynamic driving scenarios in urban settings. GPU kernels using CUDA, OptiX, and cuFFT propagate radar energy from the radar, through the scene, and back to the receiver to generate range-Doppler images displaying distance and relative velocity of surrounding objects. ANSYS will discuss the new solver and show results of automotive scenarios in busy, complex environments. The solver shows promise toward our goal of ultra-fast radar simulation.