Computer Vision

Gecko Robotics is testing Ouster's Rev8 digital lidar sensor to enhance its Cantilever platform for industrial infrastructure inspection. The Rev8's ability to output colorized 3D point clouds alongside infrared and intensity data enables Gecko's AI to more reliably detect structural anomalies that ...

Brain Corp and UC San Diego have expanded their research partnership to develop contextual 3D semantic mapping technology for autonomous robots in commercial and industrial settings. The collaboration addresses a critical gap: making next-generation AI-powered robots reliable and deployable in dynam...

Major tech companies including Hugging Face, Nvidia, and Alibaba have released open-source AI models and tools specifically for robotics, making it easier to build intelligent physical systems without specialized expertise. Hugging Face's LeRobot platform now hosts over 58,000 robotics datasets — th...

Genesis AI has launched its first foundational model, GENE-26.5, alongside human-scale robotic hands developed with China's Wuji Tech. The hands perform complex manipulation tasks including cooking, piano playing, and lab work — trained on human video data and teleoperation collected via a lightweig...

Physical Intelligence has released π0.7, a vision-language-action model that performs dexterous manipulation tasks across different robot platforms and can combine learned skills to solve new problems. The model uses multimodal prompts—language, visual subgoals, and metadata—to integrate diverse tra...

Physical Intelligence has developed RL tokens (RLT), a reinforcement learning method that allows vision-language-action models to refine precise manipulation skills using just 15 minutes of real-world robot experience. The technique adds a compressed "RL token" output to the base model, enabling a l...

NVIDIA has released the Physical AI Data Factory Blueprint, an open reference architecture that automates training data generation, augmentation, and evaluation for robotics, autonomous vehicles, and vision AI agents. The blueprint addresses a core bottleneck in Physical AI development: generating m...

Software engineer Sammy Azdoufal reverse-engineered his DJI Romo vacuum cleaner and discovered a backend vulnerability granting access to live camera feeds, microphone audio, and floor plans from nearly 7,000 devices globally. The disclosure highlights systemic security weaknesses in consumer IoT ro...

Security researcher Elise Amber Katze disclosed multiple vulnerabilities in NVIDIA's Tegra X2 SoC, including an unpatchable BootROM exploit affecting Tesla Autopilot 2 and 2.5 hardware. The research, presented at 39C3, demonstrates complete secure boot chain bypass via USB recovery mode. Since the v...

Ouster has released Rev8, the first LiDAR sensor family that captures color natively in every 3D data point, removing the need for separate camera fusion or post-processing. The flagship OS1 Max offers 500-meter range, 256 channels, and operates across extreme lighting conditions—from near-darkness ...