Pure Storage’s Post

Join us at #AutoAI as Hans Ramsl explores the top 5 MLOps challenges in autonomous driving. From data quality to real-time deployment, Hans covers strategies for scalable model training and how to ensure safety and reliability in AI systems. Register here: https://lnkd.in/dZypt8T

I guess now I can officially consider myself a total nerd! Yesterday, a scientific paper about autonomous driving made me laugh. The paper is from Waymo and introduces EMMA, a self-driving model based on Google's Gemini #LLM. EMMA combines vision and text inputs (scene descriptions) and narrates its driving decisions back step by step: https://lnkd.in/enZw9ygN Picture this: Your AI taxi driver explains: "The weather is clear, there is a pedestrian at [x,y], i better slow down because he might cross." Idk why but this scene just makes me smile. It is also pretty impressive that this approach works quite well! Here it is, a sneak peek into the future of autonomous AI Agents talking their way through solving complex tasks. And it's somehow quite funny! What do you think? #nerdhumor #autonomousdriving #aiagents

Helm.ai has launched VidGen-1, a generative AI model that creates highly realistic driving scene videos for autonomous driving development. Trained on extensive driving footage, the model uses deep neural networks and unsupervised training to produce videos with realistic appearances and consistent object motion. It can simulate various environments, weather conditions, and perspectives, helping enhance prediction tasks and generative simulations crucial for autonomous driving. Read more here: https://lnkd.in/e54eNpvM #HelmAI #VidGen1 #AutonomousDriving #AI #ADAS #DeepLearning #GenerativeAI #SelfDrivingCars #Robotics #MachineLearning #ArtificialIntelligence #TechInnovation #AutonomousVehicles #SimulationTechnology

Happy to share, that our survey paper on explainable AI for autonomous driving (XAI4AD) has just been accepted to the IEEE Transactions on Intelligent Transportation Systems! https://lnkd.in/dZ5Z8ByV Starting from existing standards and guidelines for vehicles and AI, we develop a detailed list of requirements for XAI4AD, then review the state-of-the-art using a systematic methodology. We find that XAI4AD methods can grouped into 5 main groups, the biggest of which are auxiliary explanations. We argue that these methods are insufficient to address the human-centric aspects of safe and trustworthy driving and propose a conceptual framework called SafeX that integrates the different algorithms to produce a better overall system. Finally, generative AI methods and LLMs are barely utilized in XAI4AD which could become a significant new direction in the future.

In Kodiak's latest blog, "LLMs Take the Wheel: How Generative AI Can Improve Autonomous Safety," they describe how the latest Vision-Language Models (VLMs) are enabling their self-driving trucks to navigate rare and unseen scenarios with human-like intelligence, making autonomous driving safer and more reliable. Shubham Shrivastava goes on to discuss the serious computing power required to operate VLMs in real-time, and how #Ambarella’s cutting-edge CV3-AD family of AI domain control SoCs deliver the advanced processing capabilities to bring VLMs to life. Together, we’re advancing autonomous driving with the latest generative AI technology! Read their blog, to learn more about Kodiak’s VLM implementation: https://lnkd.in/gnwXbRGY #AutonomousDriving #GenerativeAI #GenAI #LLM #VLM #AI #Automotive #AIenvisioned #KodiakRobotics

#GenerativeAI in #SelfDriving trucks from Kodiak, enabled by Ambarella Inc's CV3-AD AI domain control SoC—read their blog to learn more…

From enhancing your morning routine with #AI to powering complex applications like autonomous vehicles, the latest Welo Data guide, “Behind the Code: The Human Force Powering AI Innovation,” reveals the critical role of human touch in AI development. Discover how human intelligence powers AI effectiveness - read the guide now 🔗 https://lnkd.in/eTKF6K5x #ElevateyourData #DataLabeling #Innovation #Technology #WeloData #BusinessOutcomes Louise Law Siobhan Hanna Sarha Mavrakis Damien Norris Giancarlo Mora

🚗 The Future is Autonomous! 🚗 I recently conducted a comparative analysis of the U.S. market to highlight the impressive growth in the autonomous driving industry—visualized using Matplotlib. The importance of this technology is underscored by the substantial investments made in 2024 alone. According to the paper by Ricco Kämpfer (Team Lead Autonomous Driving Market & Strategy at P3), competition among companies in this space has reached unprecedented levels in recent years. Currently, Waymo leads the race, setting the pace for innovation and progress. As the industry evolves, it’s clear that autonomous driving will play a pivotal role in shaping the future of mobility. The link to the paper: https://lnkd.in/d9J-BGZz #AutonomousDriving #Innovation #FutureOfMobility #DataAnalysis #Waymo #AI #TechTrends

🤖 Artificial Intelligence (AI) has become an integral part of our daily lives, powering everything from virtual assistants to recommendation systems and autonomous vehicles. But how does AI actually work? Let's delve into the structural breakdown of AI to uncover its inner workings. 🚀 #ArtificialIntelligence #AI #MachineLearning #DataScience #DeepLearning #NeuralNetworks #DataPreprocessing #FeatureExtraction #ModelTraining #ModelEvaluation #Deployment #ContinuousLearning #AIInnovation #TechTrends #DataAnalytics #TechForGood #AIApplications #Innovation #TechInsights #AIInfrastructure #FutureTech #OrangeEducation

Wayve Introduces the First Vision-Language-Action Model Tested on Public Roads Wayve introduces LINGO-2, the first vision-language-action model for autonomous driving, enhancing AI explainability and human-machine interaction on roads. Read more here: https://lnkd.in/eWXt72pe