NVIDIA Research Presents Papers on Grasping, Driving, Agent Training

NVIDIA Research released three papers at CVPR on advanced robot grasping, real-time autonomous driving safety, and large-scale virtual agent training. The grasping model enables robots to adapt to new tools without prior experience. Autonomous driving systems now reason faster on actual car hardware. Virtual agents train across diverse environments before real-world deployment. These advances support scalable robotics and safer self-driving systems.