News from the AI & ML world
DeeperML - #robotics
|
staff@insidehpc.com
//
Amazon is making a substantial investment in artificial intelligence infrastructure, announcing plans to spend $10 billion in North Carolina. The investment will be used to build a cloud computing and AI campus just east of Charlotte, NC. This project is anticipated to create hundreds of good-paying jobs and provide a significant economic boost to Richmond County, positioning North Carolina as a hub for cutting-edge technology.
This investment underscores Amazon's commitment to driving innovation and advancing the future of cloud computing and AI technologies. The company plans to expand its AI data center infrastructure in North Carolina, following a trend among Big Tech companies who are building out infrastructure to meet escalating AI resource requirements. The new "innovation campus" will house data centers containing servers, storage drives, networking equipment, and other essential technology. Amazon is also focused on improving efficiency by enhancing warehouse operations through the use of AI. The company unveiled AI upgrades to boost warehouse efficiency. These upgrades center around the development of "agentic AI" robots. These robots are designed to perform a variety of tasks, from unloading trailers to retrieving repair parts and lifting heavy objects, all based on natural language instructions. The goal is to create systems that can understand and act on commands, transforming robots into multi-talented helpers, ultimately leading to faster deliveries and improved efficiency. Recommended read:
References :
@siliconangle.com
//
Hugging Face, primarily known as a platform for machine learning and AI development, is making a significant push into the robotics field with the introduction of two open-source robot designs: HopeJR and Reachy Mini. HopeJR is a full-sized humanoid robot boasting 66 degrees of freedom, while Reachy Mini is a desktop unit. The company aims to democratize robotics by offering accessible and open-source platforms for development and experimentation. These robots are intended to serve as tools for AI developers, similar to a Raspberry Pi, facilitating the testing and integration of AI applications in robotic systems. Hugging Face anticipates shipping initial units by the end of the year.
HopeJR, co-designed with French robotics company The Robot Studio, is capable of walking and manipulating objects. According to Hugging Face Principal Research Scientist Remi Cadene, it can perform 66 movements including walking. Priced around $3,000, HopeJR is positioned to compete with offerings like Unitree's G1. CEO Clem Delangue emphasized the importance of the robots being open source. He stated that this enables anyone to assemble, rebuild, and understand how they work and remain affordable, ensuring that robotics isn’t dominated by a few large corporations with black-box systems. This approach lowers the barrier to entry for researchers and developers, fostering innovation and collaboration in the robotics community. Reachy Mini is a desktop robot designed for AI application testing. Resembling a "Wall-E-esque statue bust" according to reports, Reachy Mini features a retractable neck that allows it to follow the user with its head and auditory interaction. Priced between $250 and $300, Reachy Mini is intended to be used to test AI applications before deploying them to production. Hugging Face's expansion into robotics includes the acquisition of Pollen Robotics, a company specializing in humanoid robot technology, and the release of AI models specifically designed for robots, as well as the SO-101, a 3D-printable robotic arm. Recommended read:
References :
Maximilian Schreiner@THE DECODER
//
Carnegie Mellon University has unveiled LegoGPT, a novel AI application capable of designing physically stable LEGO structures from simple text prompts. This innovative system leverages a repurposed large language model (LLM) originally created by META, combined with a custom-built math module that ensures structural integrity by accounting for gravity and other physical forces. The goal of the project was to overcome the limitations of existing 3D generative models, which often produce designs that are not feasible in the real world due to stability issues.
To train the AI, the team created a dataset of 47,000 stable LEGO structures, each accompanied by descriptive captions generated by another AI system. This extensive dataset allowed LegoGPT to learn the relationships between text descriptions and the corresponding LEGO configurations. The AI operates recursively, placing bricks one at a time and evaluating the structure's stability after each addition. If a brick introduces instability, the system employs a "rollback" feature to remove it and explore alternative placements, greatly improving the overall stability of the creations. The system is also able to produce a wide variety of stable LEGO structures and even add color and texture abilities. The system can only generate designs that fit within a 20 x 20 x 20 grid using just eight basic brick types (1 x 1, 1 x 2, 1 x 4, 1 x 6, 1 x 8, 2 x 2, 2 x 4, and 2 x 6). Researchers tested their system against other AI systems that have been built to create 3D objects and found that theirs produced a higher percentage of stable structures. The researchers have also demonstrated the practicality of LegoGPT by using robots to build the AI-designed structures, indicating its potential for automating the design and construction of stable buildings. Recommended read:
References :
Evan Ackerman@IEEE Spectrum
//
Amazon is enhancing its warehouse operations with the introduction of Vulcan, a new robot equipped with a sense of touch. This advancement is aimed at improving the efficiency of picking and handling packages within its fulfillment centers. The Vulcan robot, armed with gripping pincers, built-in conveyor belts, and a pointed probe, is designed to handle 75% of the package types encountered in Amazon's warehouses. This new capability represents a "fundamental leap forward in robotics," according to Aaron Parness, Amazon’s director of applied science, as it enables the robot to "feel" the objects it's interacting with, a feature previously unattainable for Amazon's robots.
Vulcan's sense of touch allows it to navigate the challenges of picking items from cluttered bins, mastering what some call 'bin etiquette'. Unlike older robots, which Parness describes as "numb and dumb" because of a lack of sensors, Vulcan can measure grip strength and gently push surrounding objects out of the way. This ensures that it remains below the damage threshold when handling items, a critical improvement for retrieving items from the small fabric pods Amazon uses to store inventory in fulfillment centers. These pods contain up to 10 items within compartments that are only about one foot square, posing a challenge for robots without the finesse to remove a single object without damaging others. Amazon claims that Vulcan's introduction is made possible through key advancements in robotics, engineering, and physical artificial intelligence. While the company did not specify the exact number of jobs Vulcan may create or displace, it emphasized that its robotics systems have historically led to the creation of new job categories focused on training, operating, and maintaining the robots. Vulcan, with its enhanced capabilities, is poised to significantly impact Amazon's ability to manage the 400 million SKUs at a typical fulfillment center, promising increased efficiency and reduced risk of damage to items. Recommended read:
References :
Evan Ackerman@IEEE Spectrum
//
Amazon has unveiled Vulcan, an AI-powered robot with a sense of touch, designed for use in its fulfillment centers. This groundbreaking robot represents a "fundamental leap forward in robotics," according to Amazon's director of applied science, Aaron Parness. Vulcan is equipped with sensors that allow it to "feel" the objects it is handling, enabling capabilities previously unattainable for Amazon robots. This sense of touch allows Vulcan to manipulate objects with greater dexterity and avoid damaging them or other items nearby.
Vulcan operates using "end of arm tooling" that includes force feedback sensors. These sensors enable the robot to understand how hard it is pushing or holding an object, ensuring it remains below the damage threshold. Amazon says that Vulcan can easily manipulate objects to make room for whatever it’s stowing, because it knows when it makes contact and how much force it’s applying. Vulcan helps to bridge the gap between humans and robots, bringing greater dexterity to the devices. The introduction of Vulcan addresses a significant challenge in Amazon's fulfillment centers, where the company handles a vast number of stock-keeping units (SKUs). While robots already play a crucial role in completing 75% of Amazon orders, Vulcan fills the ability gap of previous generations of robots. According to Amazon, one business per second is adopting AI, and Vulcan demonstrates the potential for AI and robotics to revolutionize warehouse operations. Amazon did not specify how many jobs the Vulcan model may create or displace. Recommended read:
References :
@developer.nvidia.com
//
References:
NVIDIA Newsroom
, NVIDIA Technical Blog
,
NVIDIA continues to advance the capabilities of AI across various sectors with its integrated hardware and software platforms. The NVIDIA Isaac GR00T N1 represents a significant leap in robotics, offering a hardware-software combination designed to provide humanoid robots with advanced cognitive abilities. This platform aims to bridge the gap between large language models and real-world dexterity, enabling robots to learn from single demonstrations and adapt to different tasks in diverse environments such as homes, factories, and disaster zones. Powered by the Jetson Thor chip and built on the Isaac robotics platform, GR00T N1 focuses on adaptability through foundation model reinforcement learning, allowing robots to reason and generalize insights, rather than simply reacting to pre-programmed instructions.
Agentic AI is transforming cybersecurity by introducing both new opportunities and challenges. These AI agents can autonomously interact with tools, environments, and sensitive data, enhancing threat detection, response, and overall security. Cybersecurity teams, often overwhelmed by talent shortages and high alert volumes, can leverage agentic AI to bolster their defenses. These systems can perceive, reason, and act autonomously to solve complex problems, serving as intelligent collaborators for cyber experts. Organizations such as Deloitte are utilizing NVIDIA AI Blueprints, NIM, and Morpheus to accelerate software patching and vulnerability management, while CrowdStrike and Trend Micro are leveraging NVIDIA AI software to improve security alert triaging and reduce alert fatigue. NVIDIA NIM Operator 2.0 enhances AI deployment by supporting NVIDIA NeMo microservices, streamlining the management of inference pipelines for MLOps and LLMOps engineers. This new release simplifies the deployment, auto-scaling, and upgrading of NIM on Kubernetes clusters, building upon the capabilities of the initial NIM Operator. The NIM Operator 2.0 introduces the ability to deploy and manage the lifecycle of NVIDIA NeMo microservices, including NeMo Customizer for fine-tuning LLMs, NeMo Evaluator for comprehensive evaluation of LLMs, and NeMo Guardrails for adding safety checks and content moderation to LLM endpoints. These enhancements provide efficient model caching and boost overall operational efficiency for deploying and managing NIM on various infrastructures, as highlighted by Cisco Systems. Recommended read:
References :
Jennifer Chu@news.mit.edu
//
References:
learn.aisingapore.org
, news.mit.edu
MIT researchers have recently made significant strides in artificial intelligence, focusing on enhancing robotics, code generation, and system optimization. One project involves a novel robotic system designed to efficiently identify and prioritize objects relevant to assisting humans. By cutting through data noise, the robot can focus on crucial features in a scene, making it ideal for collaborative environments like smart manufacturing and warehouses. This innovative approach could lead to more intuitive and safer robotic helpers in various settings.
Researchers have also developed a new method to improve the accuracy of AI-generated code in any programming language. This approach guides large language models (LLMs) to produce error-free code that adheres to the rules of the specific language being used. By allowing LLMs to focus on outputs most likely to be valid and accurate, while discarding unpromising outputs early on, the system achieves greater computational efficiency. This advancement could help non-experts control AI-generated content and enhance tools for AI-powered data analysis and scientific discovery. A new methodology for optimizing complex coordinated systems has emerged from MIT, utilizing simple diagrams to refine software optimization in deep-learning models. This diagram-based "language," rooted in category theory, simplifies the process of designing computer algorithms that control various system components. By revealing relationships between algorithms and parallelized GPU hardware, this approach makes it easier to optimize resource usage and manage the intricate interactions between different parts of a system, potentially revolutionizing the way complex systems are designed and controlled. Recommended read:
References :
@developer.nvidia.com
//
References:
blogs.nvidia.com
, developer.nvidia.com
NVIDIA Research is making significant strides in multimodal generative AI and robotics, as showcased at the International Conference on Learning Representations (ICLR) 2025 in Singapore. The company is focusing on a full-stack approach to AI development, optimizing everything from computing infrastructure to algorithms and applications. This approach supports various industries and tackles real-world challenges in areas like autonomous vehicles, healthcare, and robotics.
NVIDIA has introduced a new plug-in builder for G-Assist, which enables the integration of AI with large language models (LLMs) and various software programs. This allows users to customize NVIDIA's AI to fit their specific needs, expanding G-Assist's functionality by adding new commands and connecting external tools. These plug-ins can perform a wide range of functions, from connecting with LLMs to controlling music, and can be built using coding languages like JSON and Python. Developers can also submit their plug-ins for potential inclusion in the NVIDIA GitHub repository. NVIDIA Research is also addressing the need for adaptable robotic arms in various industries with its R²D² (Robotics Research and Development Digest) workflows and models. These innovations aim to enable robots to make decisions and adjust their behavior based on real-time data, improving flexibility, safety, and collaboration in different environments. NVIDIA is developing models and workflows for dexterous grasping and manipulation, addressing challenges like handling reflective objects and generalizing to new objects and dynamic environments. DextrAH-RGB, for example, is a workflow that performs dexterous arm-hand grasping from stereo RGB input, trained at scale in simulation using NVIDIA Isaac Lab. Recommended read:
References :
@amazon.jobs
//
References:
IEEE Spectrum
, Amazon Science homepage
,
Amazon is rapidly advancing the integration of robotics and artificial intelligence within its warehouses, marking a significant shift in how the e-commerce giant handles order fulfillment. This push for automation is not just about efficiency; it's about meeting the increasing demands of online shoppers. Amazon's fulfillment centers are becoming showcases for cutting-edge technology, demonstrating how robotics and AI can revolutionize warehouse operations. One example is the deployment of "Robin," a robotic arm capable of sorting packages for outbound shipping by moving them from conveyors to mobile robots, with over three billion successful package moves already completed across various Amazon facilities.
Amazon's robotics innovations are not limited to sorting packages. They are also focused on solving complex problems like robotic stowing, which involves intelligently placing items in cluttered storage bins. This requires robots to understand the three-dimensional world, manipulate a variety of objects, and even create space within bins by gently pushing items aside. Amazon's commitment to building safe and reliable technology that optimizes the supply chain is evident in its development of collaborative robots like Proteus, Cardinal, and Sparrow, as well as its new approach to inventory management through Containerized Storage. These systems are designed to work alongside humans safely, reducing physically demanding tasks and improving workplace ergonomics. The company has deployed more than 750,000 mobile robots across its global operations. Amazon's approach to robotics development involves rigorous testing in real-world environments, starting with small-scale implementations before wider deployment. Furthermore, Amazon is committed to upskilling its workforce. This commitment means that employees get the chance to learn new skills and use new innovative tools to deliver even more value for customers. Recommended read:
References :
@the-decoder.com
//
Hugging Face, a leading open-source AI platform, has announced its acquisition of Pollen Robotics, a French startup specializing in robotics. This strategic move aims to expand open-source robotics efforts and make robotics more accessible through transparency and community-driven development. The acquisition includes Pollen’s humanoid robot Reachy 2, which Hugging Face plans to further develop as an open-source hardware and software platform. The company believes this initiative will lower technical barriers and accelerate innovation in the field, positioning open source as a vital solution to industry challenges.
Hugging Face's CEO, Clément Delangue, emphasizes the critical importance of transparency in robotics, particularly when dealing with physical systems interacting in real-world environments. Open-source frameworks, featuring publicly available code and hardware documentation, are seen as a means to build trust and foster collaboration within the robotics community. Pollen CEO Matthieu Lapeyre views the acquisition as an opportunity to make robotics more practical and accessible, offering a transparent, community-driven alternative to the proprietary approaches often adopted by large, well-funded companies in the sector. Reachy 2 is already being utilized by various AI companies in research settings, demonstrating its capabilities in performing basic tasks such as picking up objects. Hugging Face intends to release detailed schematics, parts lists, and 3D models of Reachy 2, enabling developers to repair components or make custom modifications. By fostering an open and collaborative environment, Hugging Face hopes to counterbalance inflated expectations in the robotics sector, where public demonstrations often depict ideal scenarios, and unlock the full potential of AI-driven robotics for broader applications. Recommended read:
References :
Sam Khosroshahi@lambdalabs.com
//
References:
Fello AI
, lambdalabs.com
,
NVIDIA is pushing the boundaries of artificial intelligence in healthcare and robotics, introducing several groundbreaking advancements. One notable innovation is the DNA LLM, designed to decode the complex genetic information found in DNA, RNA, and proteins. This tool aims to transform genomic research, potentially leading to new understandings and treatments for various diseases.
The company's commitment to AI extends to robotics with the release of Isaac GR00T N1, an open-source platform for humanoid robots. This initiative is expected to accelerate innovation in the field, providing developers with the resources needed to create more advanced and capable robots. Additionally, an NVIDIA research team has developed Hymba, a family of small language models that combine transformer attention with state space models, surpassing the Llama-3.2-3B model in performance while significantly reducing cache size and increasing throughput. Recommended read:
References :
Dean Takahashi@AI News | VentureBeat
//
NVIDIA, Google DeepMind, and Disney Research are collaborating on Newton, an open-source physics engine designed to advance robot learning, enhance simulation accuracy, and facilitate the development of next-generation robotic characters. Newton is built on NVIDIA’s Warp framework and aims to provide a scalable, high-performance simulation environment optimized for AI-driven humanoid robots. MuJoCo-Warp, a collaboration with Google DeepMind, accelerates robotics workloads by over 70x, while Disney plans to integrate Newton into its robotic character platform for expressive, interactive robots.
The engine's creation is intended to bridge the gap between simulation and real-world robotics. NVIDIA will also supercharge humanoid robot development with the Isaac GR00T N1 foundation model for human-like reasoning. Newton is built on NVIDIA Warp, a CUDA-based acceleration library that enables GPU-powered physics simulations. Newton is also optimized for robot learning frameworks, including MuJoCo Playground and NVIDIA Isaac Lab, making it an essential tool for developers working on generalist humanoid robots. This initiative is part of NVIDIA's broader effort to accelerate physical AI progress. Recommended read:
References :
Michael Nuñez@venturebeat.com
//
References:
venturebeat.com
, AIwire
,
Nvidia has made significant strides in enhancing robot training and AI capabilities, unveiling innovative solutions at its GTC conference. A key announcement was Cosmos-Transfer1, a groundbreaking AI model designed to generate photorealistic simulations for training robots and autonomous vehicles. This model bridges the gap between virtual training environments and real-world applications by using multimodal inputs to create highly realistic simulations. This adaptive multimodal control system allows developers to weight different visual inputs, such as depth or object boundaries, to improve the realism and utility of the generated environments.
Nvidia also introduced its next-generation GPU superchips, including the second generation of the Grace Blackwell chip and the Vera Rubin, expected in the second half of 2026. The Vera Rubin will feature 288GB of high-bandwidth memory 4 (HBM4) and will be paired with CPUs boasting 88 custom Arm cores. These new chips promise substantial increases in compute capacity, with Rubin delivering a 900x speedup compared to the previous generation Hopper chips. This positions Nvidia to tackle the increasing demands of generative AI workloads, including training massive AI models and running inference workloads. Recommended read:
References :
Shelly Palmer@Shelly Palmer
//
References:
Shelly Palmer
, Upward Dynamism
AI agents are rapidly evolving, becoming a focal point in the tech industry. OpenAI has released an API designed to make the deployment of AI agents significantly easier, opening up new possibilities for developers. These agents are capable of tasks like searching the web in real-time and analyzing extensive datasets, heralding a new era of automation and productivity. Experts predict substantial growth in the AI agent market, forecasting an increase from $5 billion in 2024 to over $47 billion by 2030.
AI agents are being designed as smart software programs that can understand and act independently to achieve assigned goals. Unlike chatbots or rigid scripts, AI agents break down complex tasks into manageable steps, adapt their strategies as needed, and learn from their experiences. Core elements include tasks, models (like GPT or Claude), memory for storing context, and external tools such as APIs and web access. These components work together iteratively, planning, acting, and adjusting as necessary without constant human intervention. Recommended read:
References :
@Google DeepMind Blog
//
Google is pushing the boundaries of AI and robotics with its Gemini AI models. Gemini Robotics, an advanced vision-language-action model, now enables robots to perform physical tasks with improved generalization, adaptability, and dexterity. This model interprets and acts on text, voice, and image data, showcasing Google's advancements in integrating AI for practical applications. Furthermore, the development of Gemini Robotics-ER, which incorporates embodied reasoning capabilities, signifies another step toward smarter, more adaptable robots.
Google's approach to robotics emphasizes safety, employing both physical and semantic safety systems. The company is inviting filmmakers and creators to experiment with the model to improve the design and development. Veo builds on years of generative video model work, including Generative Query Network(GQN),DVD-GAN,Imagen-Video,Phenaki,WALT,VideoPoetandLumiere— combining architecture, scaling laws and other novel techniques to improve quality and output resolution. Recommended read:
References :
Tris Warkentin@The Official Google Blog
//
Google AI has released Gemma 3, a new family of open-source AI models designed for efficient and on-device AI applications. Gemma 3 models are built with technology similar to Gemini 2.0, intended to run efficiently on a single GPU or TPU. The models are available in various sizes: 1B, 4B, 12B, and 27B parameters, with options for both pre-trained and instruction-tuned variants, allowing users to select the model that best fits their hardware and specific application needs.
Gemma 3 offers practical advantages including efficiency and portability. For example, the 27B version has demonstrated robust performance in evaluations while still being capable of running on a single GPU. The 4B, 12B, and 27B models are capable of processing both text and images, and supports more than 140 languages. The models have a context window of 128,000 tokens, making them well suited for tasks that require processing large amounts of information. Google has built safety protocols into Gemma 3, including a safety checker for images called ShieldGemma 2. Recommended read:
References :
|
BenchmarksBlogsResearch Tools |