NVIDIA-ISAAC-ROS/isaac_ros_visual_slam

isaac_ros_visual_slam is a cornerstone of the NVIDIA Isaac ROS ecosystem. Its primary moat is not just the code, but the tight integration with NVIDIA's proprietary cuVSLAM library and CUDA-accelerated kernels. While the ROS 2 wrapper is open, the core performance-critical logic is a 'black box' optimized specifically for NVIDIA silicon (Jetson and dGPUs). With over 1,300 stars and significant longevity (4.5 years), it has become the de facto standard for high-performance visual SLAM on autonomous mobile robots (AMRs) using ROS 2. Competitors include open-source projects like ORB-SLAM3 and VINS-Mono, which offer high accuracy but often struggle with real-time performance on edge hardware without extreme optimization. Commercial competitors like SLAMcore offer similar performance but lack the same level of free integration with the dominant hardware platform. The 'platform domination risk' is high because NVIDIA effectively uses this software to create hardware lock-in; moving away from Isaac ROS often means a significant drop in compute efficiency or a move to significantly more expensive hardware. Frontier labs (OpenAI/Anthropic) are currently focused on high-level reasoning and are unlikely to compete in the low-level, real-time spatial computing layer required for robotics in the near term.