Sony’s Post

𝗟𝗲𝘃𝗲𝗹 𝘂𝗽 𝘆𝗼𝘂𝗿 𝘃𝗶𝗿𝘁𝘂𝗮𝗹 𝘄𝗼𝗿𝗹𝗱𝘀 𝘄𝗶𝘁𝗵 𝗡𝗩𝗜𝗗𝗜𝗔'𝘀 𝗧𝗲𝗦𝗠𝗼! This innovative method empowers you to create text-controlled, scene-aware motion generation. Imagine crafting realistic and diverse human-object interactions for your projects – from navigation and sitting to specific orientations, body positions, and poses. TeSMo opens doors to a whole new level of detail and dynamism in virtual worlds. #NVIDIA #TeSMo #AI #motiongeneration #virtualreality #3D NVIDIA

🚀 NVIDIA has launched its most ambitious project yet - GR00T! This general-purpose foundation model promises to revolutionise humanoid robot learning both in simulation and in the real world. GR00T enables humanoid embodiments to learn from a small number of human demonstrations with imitation learning and NVIDIA Isaac Lab for reinforcement learning, as well as generating robot movements from video data. 🤖 With the ability to take multimodal instructions and past interactions as input, GR00T produces the actions for the robot to execute. Embodied AI and robotics will be the ultimate gameplay for Nvidia, opening up another huge market opportunity. 🌟 Join the revolution with #NVIDIA AI, #artificialintelligence, #chatgpt, #innovation, #technology.

Just announced📢📢 project! GROOT! NVIDIA's most ambitiou GROOT🫚🫚 is a general-purpose foundation model that promises to transform humanoid robot learning in simulation and the real world. Trained in NVIDIA GPU-accelerated simulation, GROOT enables humanoid embodiments to learn from a handful of human demonstrations with imitation learning and NVIDIA Isaac Lab for reinforcement learning, as well as generating robot movements from video data. The GROOT model takes multimodal instructions and past interactions as input and produces the actions for the robot to execute. Embodied Al and robotics will be the ultimate gameplay for Nvidia to open another huge market opportunity. Super excited about this, I need one ASAP NVIDIA #nvidiapartner #artificialintelligence #chatgpt #innovation #technology

We're thrilled to share insights from Deepu Talla's remarkable keynote at #CES2024, where NVIDIA and its partners showcased groundbreaking advancements in #AI and #robotics! 🤖 Talla emphasized the significance of GPU-accelerated large language models (#LLMs) as the cornerstone of robotic intelligence, enabling more natural interactions with humans. 🧠🤝 A key highlight is the dual-computer model, where the "AI factory" creates and refines AI models, while the runtime environment executes them in robots. This dual-brain system, coupled with NVIDIA's data center compute infrastructure, facilitates the simulation and training of AI models. 💻 Talla also underscored the role of LLMs in dismantling technical barriers, enabling users to interact with #robots through natural language. The NVIDIA #Picasso tool, for instance, transforms users into technical artists capable of designing intricate robotics workcells. 🎨 Generative AI's impact is already felt across industries. NVIDIA #Isaac and #Jetson platforms empower over 1.2 million developers, along with 10,000 customers and partners, in shaping the landscape of AI-powered robotics. 🌐🤖 For more profound insights:  https://lnkd.in/dQjb-gSf #CES2024 #Innovation #NVIDIA #DeepuTalla #TechAdvancements NVIDIA Robotics #GenerativeAI #NVIDIAOmniverse

Have you seen Motorola's new video for the Razr50? It's generating a lot of buzz because it was made entirely with... yes, you guessed it, artificial intelligence. Here’s my perspective as a marketer, but I’d love to hear your thoughts too: 1. The news is that Motorola France’s "Styled By Moto" video was created using artificial intelligence, making it one of the first advertisements ( Toys"R"Us released one a few days ago) produced with this technology. This is still a very valid tactic for generating visibility and building innovation as an attribute. 2. However, the piece doesn't seem to fully leverage the benefits of this new technology, which promises to generate content almost instantly from a prompt. According to The Rundown AI, producing the video took 4 months and involved the use of at least 7 tools (Sora, Adobe Firefly, Midjourney, Krea, Magnific, Luma, Udio for audio...). 3. What's interesting is that it appears to be a video where the idea (a fashion piece showcasing models in outfits inspired by the Motorola logo) is driven by the technology, when in marketing, the idea should always come before the technology. In summary, I think being a pioneer in AI is a smart move for Motorola, as it’s generating visibility and conferring positive attributes. However, I also believe this tactic will wear out quickly. Above all, it’s another indication that the industry is changing irreversibly. #Marketing #AI #ArtificialIntelligence #DigitalMarketing #Advertising #BrandStrategy #VideoMarketing #CreativeTech #MarketingInsights #FutureOfMarketing https://lnkd.in/d32ZrENE

I've been experimenting with various generative AI models for just over a year now. The idea of taking in-game animated characters, and transforming them into photo realistic avatars is intriguing. As we move into a new era of frame generation with NVIDIA DLSS 3.5 and AMD FSR 3 - the future will be a combination of rendering approximations (low poly) of what must be displayed and then have AI models build the final frame from that in realtime. Here are some little experiments of avatar replacement with Alfira at The Last Light Inn (Baldurs Gate 3 from Larian Studios) - even minor adjustments can made a significant different. Rendered from a single frame at what would be roughly 0.3fps, but we'll get there eventually. Full res screenshot: http://bit.ly/3LEoBb9

NVIDIA's Game-Changer: GR00T! 🤖 NVIDIA has just unveiled its most ambitious project yet: GR00T. This groundbreaking foundation model is set to revolutionize humanoid robot learning, both in simulation and the real world. Trained extensively in GPU-accelerated simulation, GR00T empowers humanoid embodiments to learn from a mere handful of human demonstrations. Leveraging imitation learning and NVIDIA Isaac Lab for reinforcement learning, this groundbreaking model can even generate robot movements from video data. At its core, the GR00T model processes multimodal instructions and past interactions to produce precise actions for the robot to execute. With its capabilities, embodied AI and robotics are poised to become the ultimate gameplay, opening up vast market opportunities. For more insights into cutting-edge technologies like GR00T, follow me Dr. Biji Kumar R. Disclaimer: This post is for informational purposes only. All credits for the video go to the respective creator. #gr00t #embodiedai #roboticsrevolution #artificialintelligence

Good afternoon, is summer over? Fourier Image Lab now has comprehensive training course on imaging for automotive and robotic cameras and perception aggregating educational materials that I've published during last few years at conferences such as AutoSens, Auto.Ai, Edge AI and Vision Alliance and others. You can learn about fundamentals of automotive imaging, driving exposure and gain control, signal and noise, color and different CFAs, using augmentation policies to collect training data, adaptation of conventional ISP for computer vision, wide angle cameras, calibration and distortion, developing flicker mitigated data driven cameras and custom lens design, limitations of vision-only approach, feasibility study of migrating to a high resolution sensor, detection probabilities for different objects, ISP tuning and practical aspects of measuring key image quality KPIs, high level design, architecture and infra of automotive SOC and much more... Sincere thanks to my managers, mentors and colleagues Yimeng Zhang Brijesh Tripathi Atif Sarwari Jeong Ho Lyu Norman Koren Boyd Fowler Robin Jenkin James Tornes Ross Jatou Nitin Sampat Alexis Lluis Alexander Zyazin Jim Keller Jeff Bier Robert Stead Alina Kuteeva Nicoleta Kallée Alexey Lukin Andrey Tovchigrechko An Tran Yi Zhang Dion Tims Hao Guo Yifei Wu Philip Mao Kai Chen Yiming Liu Bill Pu Jie Mao Kevin Sheu Clay Yu Irene (Shuohan) Fu Amit Mehta Roark Raleigh Tsuyoshi Hara Amit Mehta Alexander Braun and everyone else who made this work possible, organized conferences and motivated and supported the efforts. Some materials are posted online here for the first time, the knowledge base will grow. Not hosted at Google and visible in China. Repost would be appreciated! #autonomousdriving #robotics #education #camera #imagequality #perception #computervision #machinelearning #ai #edgeai #isptuning

I am really excited, yet not surprised, to see this. AI is the brain behind robots, yet numerous challenges still exist in robotics. However, with further advancements in AI, increased computing power, and new-wave of computing paradigms like neuromorphic and quantum computing, we will discover the materials and achieve the dexterity required to build a complete anthropomorphic humanoid.