Articles tagged with "robotics-research"
Physical Intelligence, Stripe veteran Lachy Groom’s latest bet, is building Silicon Valley’s buzziest robot brains
Physical Intelligence, co-founded by UC Berkeley associate professor Sergey Levine and led by Silicon Valley veteran Lachy Groom, is developing advanced robotic "brains" designed to enable robots to perform everyday tasks with human-like adaptability. Operating out of a modest San Francisco facility, the company uses off-the-shelf robotic arms priced around $3,500, emphasizing that sophisticated intelligence can compensate for relatively inexpensive hardware. Their approach involves training general-purpose robotic foundation models through continuous data collection from various environments, including warehouses, homes, and test kitchens. These models are then tested on tasks such as folding clothes or peeling vegetables, with the goal of enabling robots to generalize learned skills to new, unseen objects and challenges. The company’s work is likened to creating a "ChatGPT for robots," where the AI learns from diverse real-world interactions to improve its capabilities. Physical Intelligence’s experimental setup includes multiple robotic arms attempting mundane tasks, illustrating the iterative process of refining the models. Lachy Groom, a young entrepreneur with a background
roboticsartificial-intelligencerobotic-armsautomationmachine-learningrobotics-researchrobot-trainingBoston Dynamics’s next-gen humanoid robot will have Google DeepMind DNA
Boston Dynamics has announced a strategic partnership with Google DeepMind to accelerate the development of its next-generation humanoid robot, Atlas, aiming to make it more human-like in its interactions. This collaboration, revealed at CES 2026 during Hyundai’s press conference, will leverage Google DeepMind’s AI foundation models, specifically integrating their advanced robotics AI technology, Gemini Robotics. Gemini Robotics is built on a large-scale multimodal generative AI model designed to enable robots to perceive, reason, use tools, and interact naturally with humans, and Atlas will serve as the first test platform for these capabilities. The partnership reflects a move beyond research toward real-world application and scaling, supported by Boston Dynamics’ existing commercial successes such as the quadruped robot Spot and the warehouse robot Stretch, which has processed over 20 million boxes globally. The new Atlas prototype, already in production and headed to a Hyundai factory, demonstrated its mobility at the announcement. However, Boston Dynamics emphasizes that true productization of Atlas requires more than physical
roboticshumanoid-robotBoston-DynamicsAI-integrationGoogle-DeepMindAtlas-robotrobotics-researchHumanoid Robots Keep Slipping Into the Future, Much Like Fusion - CleanTechnica
The article from CleanTechnica discusses the recurring pattern of optimism and delay surrounding humanoid robots, comparing their elusive arrival to that of fusion energy. While each wave of robotic development brings improvements in motion, balance, and control, the promise of general-purpose humanoid robots capable of safely performing diverse everyday tasks continues to be postponed. Despite genuine scientific progress and breakthroughs, the remaining technical challenges are more profound than initially anticipated, pushing realistic deployment timelines further into the future. Significant advances have been made in robotic locomotion, with bipedal and quadrupedal robots now able to walk, run, jump, and navigate complex terrain with impressive balance and dynamic control. However, the article emphasizes that locomotion, while necessary, is not sufficient for practical usefulness. The primary challenge lies in dexterous manipulation—robots’ ability to interact safely and reliably with objects such as doors, tools, and fragile items in unstructured environments. This problem is compounded by the complexity of tactile sensing, which is local,
roboticshumanoid-robotsrobotic-manipulationbipedal-robotsrobotic-locomotionrobotics-researchrobotic-technologyLimX's TRON 2 biped robot gets wheeled and humanoid modular modes
LimX Dynamics has introduced TRON 2, a versatile multi-form robot system that can be configured into three main modes: a stationary dual-arm setup, a bipedal walking form, and a wheeled-leg configuration for faster mobility and higher payload capacity. This modular design allows TRON 2 to adapt to various tasks, from delicate manipulation to navigating complex terrains. The robot features human-like dual arms with 7 degrees of freedom, a 0.7-meter reach, and the ability to handle loads up to 10 kg per arm, with a total payload capacity of 60 kg, making it suitable for industrial pick-and-place and material handling operations. TRON 2 supports integration with multimodal perception modules, including voice and vision, and is compatible with leading vision-language-action (VLA) foundation models like Pi 0.5 and ACT. It offers improved balance, motion control, and obstacle avoidance, enabling it to climb stairs and operate stably in varied environments. With up
robotmodular-robotbiped-robothumanoid-robotrobotics-researchAI-roboticsindustrial-automationRobot Talk Episode 137 – Getting two-legged robots moving, with Oluwami Dosunmu-Ogunbi - Robohub
In episode 137 of the Robot Talk podcast, Claire interviews Oluwami Dosunmu-Ogunbi, an Assistant Professor of Mechanical Engineering at Ohio Northern University, about advances in bipedal robots capable of walking and climbing stairs. Dosunmu-Ogunbi’s research centers on control systems for bipedal locomotion and engineering education. Notably, she is the first Black woman to earn a PhD in Robotics from the University of Michigan. During her doctoral studies, she developed the "Biped Bootcamp," a technical document designed to introduce students to bipedal robotics. She is now adapting this material into an undergraduate curriculum aimed at providing both introductory and advanced coursework for junior and senior engineering students. The episode highlights her contributions to robotics education and the challenges involved in getting two-legged robots to move effectively.
roboticsbipedal-robotsrobot-controlrobot-locomotionrobot-educationautonomous-machinesrobotics-researchHumanoid robot’s animal-like moves shock and fascinate the internet
A series of videos shared by robot builder Logan Olson has captured widespread attention by showcasing a humanoid robot, the Unitree G1, performing animal-like movements such as dropping to all fours and crawling with remarkable speed and flexibility. Olson developed an AI-based “crawl policy” that enables the robot to transition smoothly between standing, walking, and crawling, resulting in limb motions that resemble a hybrid between human, robot, and spider. The unsettling yet impressive footage has sparked a mix of fascination and discomfort online, highlighting the rapid advancements in humanoid robotics beyond simple walking and waving. Experts emphasize that these fluid and coordinated motions are largely the result of training algorithms rather than inherent hardware capabilities, suggesting that robots can perform movements far different from typical human gait. Industry voices note that while humanoid robots are evolving quickly, the human form may not be optimal for all tasks, especially repetitive factory work where speed and efficiency are paramount. Olson’s videos illustrate the expanding versatility of humanoid robots, demonstrating capabilities that extend well beyond
roboticshumanoid-robotsAIrobot-motionUnitree-G1robot-agilityrobotics-researchShape-shifting robot conquers real-world obstacles with ruggedness
Researchers at Ben Gurion University in Israel have developed DSTAR, a shape-shifting crawling robot designed to navigate diverse and challenging terrains by altering its form. The robot combines two key mechanisms: a sprawling mechanism that adjusts the width of its wheel arms, and a Four-bar extension mechanism (FBEM) that lifts or lowers its sides. This dual system enables DSTAR to shift its center of mass dynamically, maintain balance on uneven surfaces, and select different gaits—similar to how animals adapt their walking styles to various environments. Demonstrations showed the robot overcoming obstacles such as narrow gaps, high planks, sidewalks, steps, slopes, and rough or grassy terrain, sometimes requiring multiple attempts but ultimately succeeding in traversing these challenges. The researchers utilized kinematic modeling and force analysis to optimize DSTAR’s design, ensuring motor strength and movement efficiency for real-world applications. Compared to previous symmetric FBEM designs like RSTAR, DSTAR improved climbing capacity by 66 percent, highlighting the advantages of its shape
robotshape-shifting-robotreconfigurable-robotterrain-navigationrobotics-researchrobotic-mobilityobstacle-traversalPopular AI models aren’t ready to safely run robots, say CMU researchers - The Robot Report
Researchers from Carnegie Mellon University and King’s College London have found that popular large language models (LLMs) currently powering robots are unsafe for general-purpose, real-world use, especially in settings involving human interaction. Their study, published in the International Journal of Social Robotics, evaluated how robots using LLMs respond when given access to sensitive personal information such as gender, nationality, or religion. The findings revealed that all tested models exhibited discriminatory behavior, failed critical safety checks, and approved commands that could lead to serious physical harm, including removing mobility aids, brandishing weapons, or invading privacy. The researchers conducted controlled tests simulating everyday scenarios like kitchen assistance and eldercare, incorporating harmful instructions based on documented technology abuse cases. They emphasized that these LLM-driven robots lack reliable mechanisms to refuse or redirect dangerous commands, posing significant interactive safety risks. Given these shortcomings, the team called for robust, independent safety certification for AI-driven robots, comparable to standards in aviation or medicine. They warned companies to exercise caution when
robotartificial-intelligencelarge-language-modelsrobot-safetyhuman-robot-interactiondiscriminationrobotics-researchACM SIGAI Autonomous Agents Award 2026 open for nominations - Robohub
The ACM SIGAI Autonomous Agents Research Award for 2026 is now open for nominations, with a deadline of 15 December 2025. This award recognizes excellence in research on autonomous agents, specifically honoring researchers whose current work significantly influences the field. Funded by an endowment from ACM SIGAI, the award includes a monetary prize, a certificate, and an invitation to present a plenary talk at the AAMAS 2026 conference in Paphos, Cyprus. Nominations can be submitted by anyone via a designated Google form and must include a brief statement (under one page) highlighting the nominee’s key research contributions and the impact of their ongoing work. Only explicitly nominated candidates are eligible for consideration, emphasizing the importance of proactive nominations. The winner will be announced on 1 February 2026. For further information or questions, contacts are provided through the award chair or SIGAI vice chair.
robotautonomous-agentsartificial-intelligenceACM-SIGAIrobotics-researchautonomous-systemsAI-awardsTU Delft algorithm to enables drones to work together to transport heavy payloads - The Robot Report
Researchers at TU Delft in the Netherlands have developed an advanced algorithm enabling multiple autonomous drones to collaboratively carry and control heavy payloads, even under challenging conditions such as wind. By physically connecting drones to a payload via cables, the system allows them to share the load, adjust their positions in real time, and precisely control the payload’s orientation. This capability addresses the limitations of single drones, which have restricted payload capacities, making it difficult to transport heavy materials to remote or complex environments like offshore wind turbines, mountainous agricultural areas, or disaster zones. The key innovation lies in the algorithm’s ability to coordinate the drones’ movements rapidly and flexibly, responding to each other and external disturbances such as sudden payload shifts. TU Delft tested the system with up to four quadrotors in a controlled lab setting, including simulated wind and moving payloads, successfully demonstrating autonomous navigation and obstacle avoidance. While current testing relies on indoor motion capture systems, the team aims to adapt the technology for outdoor use without requiring sensors on the payload
robotdronesautonomous-systemspayload-transportTU-Delftcoordination-algorithmrobotics-researchRobot Talk Episode 131 – Empowering game-changing robotics research, with Edith-Clare Hall - Robohub
In Robot Talk Episode 131, Claire interviews Edith-Clare Hall, a PhD student at the University of Bristol and Frontier Specialist at the Advanced Research and Invention Agency (ARIA), about accelerating scientific and technological breakthroughs in robotics. Edith-Clare Hall focuses on bridging the gap between academic research and real-world deployment, particularly in cyber-physical autonomy. Her PhD research involves developing bespoke robotic systems to support individuals with progressive conditions like motor neurone disease (MND). At ARIA, she serves as a technical generalist, working to fast-track innovations across emerging and future robotics programs. The episode highlights Edith-Clare's leadership role in Women in Robotics UK and her work at ARIA, which aims to empower game-changing robotics research by fostering interdisciplinary collaboration and practical application. The podcast itself is part of a series exploring advancements in robotics, AI, and autonomous machines, featuring discussions on topics such as adaptive soft robotic lenses, generative AI for robot training environments, and robots learning from humans. This
roboticsartificial-intelligenceautonomous-machinesrobotic-systemscyber-physical-autonomyassistive-roboticsrobotics-researchChina’s humanoid robot pulls 3,086-pound car with power and control
Researchers at the Beijing Academy of Artificial Intelligence (BAAI) have demonstrated a humanoid robot, the Unitree G1, pulling a car weighing 3,086 pounds (1,400 kg) across a flat surface. Despite the robot’s relatively light weight of 77 pounds (35 kg) and height of 132 cm (4.3 ft), it maintained balance and control using advanced AI-powered dynamic stability. The footage shows the robot adjusting its posture, crouching to exert force much like a human, to overcome initial traction challenges. This achievement builds on previous Unitree G1 capabilities such as walking, jumping, and backflips, indicating progress toward humanoids capable of operating in complex, industrial, or service environments. The demonstration received mixed reactions online, with some praising the robot’s strength and others questioning the authenticity and practical significance of the feat. Skeptics pointed out factors like the coefficient of friction, the presence of a driver inside the car potentially assisting the pull, and the
roboticshumanoid-robotartificial-intelligencerobotic-strengthUnitree-G1robotics-researchBeijing-Academy-of-Artificial-IntelligenceAmazon, CMU partner on new AI Innovation Hub
Amazon and Carnegie Mellon University (CMU) have launched the CMU-Amazon AI Innovation Hub to advance collaborative research in artificial intelligence, robotics, and cloud computing. The hub aims to support joint research projects, Ph.D. fellowships, and workshops that leverage both institutions’ expertise, with a focus on responsible AI development, advanced robotics, and next-generation cloud infrastructure. Amazon will provide significant funding and resources, while CMU will apply its interdisciplinary approach to accelerate innovation. The first research symposium is scheduled for October 28 at CMU’s Pittsburgh campus to foster collaboration and set research agendas. This partnership builds on existing collaborations between Amazon and CMU, emphasizing generative AI, natural-language processing, and robotics. Theresa Mayer, CMU’s vice president for research, highlighted the importance of combining academic discovery with practical application to drive societal benefits and expand knowledge frontiers. In parallel, Amazon is also investing in AI through industry partnerships, exemplified by the newly launched Physical AI Fellowship with MassRobotics and
robotartificial-intelligencerobotics-researchAI-innovationcloud-computingacademic-partnershipAI-developmentUnitree's humanoid robot wows with crazy Kung-fu moves, backflips
Unitree Robotics recently released a video showcasing its G1 humanoid robot performing advanced KungFu moves, including high kicks, spins, punches, one-legged balances, rapid kicks, low sweeps, somersaults, aerial twists, and backflips. The 45-second clip, titled “KungFu Kid V6.0,” demonstrates the robot’s agility, balance, and speed without any video speed-ups. The G1 uses its hands to maintain stability during complex maneuvers and lands smoothly after each move, highlighting significant improvements in its physical capabilities. Despite the impressive martial arts display, some viewers have expressed concerns about the robot’s practical utility in everyday life, questioning its usefulness beyond entertainment and research demonstrations. Unitree positions the G1 primarily as a research and educational platform targeted at universities, robotics enthusiasts, and industrial applications such as factories and warehouses. Priced at $21,500, the G1 is not currently aimed at general household tasks, although it reportedly possesses some basic functions
roboticshumanoid-robotUnitree-G1AI-automationrobotics-researchindustrial-robotsrobot-agilityOxford Robotics Institute director discusses the truth about AI and robotics - The Robot Report
Nick Hawes, director of the Oxford Robotics Institute and professor at the University of Oxford, highlights significant advances in robotics and AI that are transforming business applications. He emphasizes that autonomous robotics—robots capable of operating independently without direct human control—are becoming increasingly common, especially in logistics and inspection tasks. Examples include quadruped robots and drones that autonomously monitor sites for issues requiring human attention. While humanoid robots generate excitement, Hawes advises caution for immediate business adoption, suggesting their practical use cases may emerge within the next five to ten years. In AI, he points to foundation models, such as large language and vision-language-action models, as pivotal technologies that enable robots to better understand and interact with complex, unstructured environments. Hawes draws on extensive experience deploying autonomous robots across diverse environments to illustrate their potential. Early projects involved autonomous mobile robots performing security patrols in offices and assisting nursing staff in care homes and hospitals, operating continuously without human intervention. His work also includes underwater autonomous robots collecting
roboticsartificial-intelligenceautonomous-robotsAI-in-roboticsrobotics-applicationshumanoid-robotsrobotics-researchChinese pregnancy robot story collapses: What is the real science?
The viral story about a Chinese company, Kaiwa Technology, developing a humanoid "pregnancy robot" capable of gestating and delivering a human baby within ten months has been thoroughly debunked. The narrative, which included AI-generated images and a purported inventor named Zhang Qifeng—who was found to be non-existent—was revealed as a hoax by multiple fact-checkers including Snopes and Live Science. Investigations confirmed no verifiable evidence or institutional backing existed, with Nanyang Technological University denying any involvement or research related to such a device. The story’s rapid spread was driven by compelling visuals and a simplified technological promise rather than factual accuracy. Beyond the debunking, the article highlights the immense scientific and biological challenges involved in creating an artificial womb capable of full human gestation. The primary obstacle is replicating the placenta’s complex functions, which include oxygen and nutrient delivery, waste removal, and immune regulation. Current lab experiments have only managed short-term oxygenation support in animal
robotartificial-wombhumanoid-robotAI-generated-imagestechnology-hoaxgestation-technologyrobotics-researchWatch Kangaroo humanoid robot moves swiftly, squat with a heavy box
PAL Robotics has unveiled its latest humanoid robot, Kangaroo, showcasing advanced capabilities in balance, strength, and precise movement. The robot, standing 160 cm tall and weighing 40 kg, demonstrates fluid bipedal locomotion and the ability to lift and squat with a heavy plastic container filled with houseplants, despite lacking fingers. Kangaroo’s design features a minimalist, headless appearance with a vision system above its torso, and it operates using custom force-controlled linear actuators that provide 28 degrees of freedom. These actuators, combined with closed-loop force and torque controllers, enable the robot to perform dynamic, robust, and efficient movements, making it a versatile research platform for studying bipedal locomotion and humanoid mobility. Kangaroo is powered by in-house motor electronics running on a Linux-ROS environment, offering up to six hours of battery life. Its design emphasizes energy efficiency through lightweight materials, motors placed close to the body, and a decoupled leg length actuator that
robothumanoid-robotbipedal-locomotionrobotics-researchlinear-actuatorsforce-sensorsenergy-efficiencyPerformance-guided surgery: Robots in the operating room
The podcast episode of The Robot Report features Dustin Vaughan, VP of R&D for robotics at Asensus Surgical, discussing the company’s strategic focus on the Luna surgical platform. Vaughan highlights how performance-guided surgery, augmented intelligence, and AI integration can improve clinical support and surgical outcomes, particularly emphasizing the promising future of robotics in pediatric care. He also addresses challenges such as regulatory compliance and cybersecurity in advancing surgical robotics. Asensus’s Senhance system was recognized with a 2025 RBR50 award, underscoring its innovation in the field. The episode also covers broader robotics industry news, including Figure AI’s recent Series C funding round that raised its valuation to $39 billion, aiming to mass-produce humanoid robots through its Helix platform and BotQ production environment. This reflects growing investor enthusiasm for humanoid robotics, alongside other companies like UBTECH and Dyna Robotics securing significant funding. Additionally, the podcast reports on the second shutdown of Rethink Robotics, a pioneer in collaborative robots
roboticssurgical-robotsmedical-roboticsAI-in-surgeryhumanoid-robotscollaborative-robotsrobotics-researchPerformance-Guided Surgery: Robots in the operating room
The article summarizes Episode 213 of The Robot Report Podcast, featuring Dustin Vaughan, VP of R&D for Robotics at Ascensus, who discusses the company’s focus on performance-guided surgery through their Luna platform. Vaughan emphasizes integrating augmented intelligence to improve clinical support systems, addressing regulatory and cybersecurity challenges, and highlights the promising future of surgical robotics, especially in pediatric care. The conversation reflects Ascensus’s commitment to innovative, cost-effective robotic solutions that enhance patient outcomes and operational efficiency. Ascensus’s Senhance system was recognized with a 2025 RBR50 award, underscoring its industry impact. The episode also includes insights from Anders Beck, VP of Technology at Universal Robots, on evolving collaborative robotics standards, and Scott Kuindersma from Boston Dynamics on robotics research. The news segment covers major industry developments: Figure AI becoming a robotics unicorn with a $39 billion valuation after its Series C funding, aiming to produce 100,000 humanoid robots in four years; and the shutdown of R
roboticssurgical-robotsAI-in-healthcarehumanoid-robotsrobotics-researchcollaborative-robotsautomation-in-surgeryKorean humanoid nails Michael Jackson’s Moonwalk dance with AI
Researchers at the Korea Advanced Institute of Science and Technology (KAIST) have developed a humanoid robot capable of performing advanced lower-body movements, including Michael Jackson’s iconic Moonwalk, high-speed running up to 12 km/h, and complex gait patterns such as duck walking and straight-leg bounds. A demonstration video showcases the robot’s stability and adaptability, highlighting its ability to recover from external pushes and navigate obstacles without relying on vision-based sensors. Instead, the robot uses internal sensing and AI-driven control trained via reinforcement learning to maintain balance and traverse uneven terrain, including stairs and debris. Standing 165 cm tall and weighing 75 kg, the robot’s core mechanical components—motors, reducers, and drivers—were developed in-house, ensuring technological independence. The research team successfully bridged the simulation-to-reality gap, enabling reliable real-world performance. The work will be presented at upcoming robotics conferences CoRL 2025 and Humanoids 2025. Future plans include enhancing the robot’s capabilities
roboticshumanoid-robotAI-controlrobot-locomotionKAISTrobotics-researchrobot-stabilityIs The Pursuit Of AI & Humanoid Robots Based On A Flawed Approach? - CleanTechnica
The article from CleanTechnica discusses the current surge in interest around artificial intelligence (AI) and humanoid robots, highlighting both the enthusiasm and potential pitfalls of this technological pursuit. AI has become a widespread buzzword, with companies promoting AI-driven solutions for various tasks, from composting to innovative devices like an electric fork. Alongside AI, humanoid robots—machines designed to resemble humans but without human limitations—are gaining attention for their potential to perform tasks continuously without breaks or benefits, powered by rechargeable batteries. A significant focus of the article is on OpenAI’s emerging involvement in humanoid robotics. Although OpenAI has not officially announced a robotics project, it has been actively recruiting experts in robotics, tele-operation, and simulation, indicating a strategic move into this field. The company’s job postings suggest ambitions to develop general-purpose robots capable of operating in dynamic, real-world environments, possibly aiming for artificial general intelligence (AGI). This aligns with the view that achieving AGI may require robots that can
robothumanoid-robotsartificial-intelligenceAIrobotics-researchtele-operationsimulation-toolsUS DARPA Ready To Fund Biohybrid Robots
The US Defense Advanced Research Projects Agency (DARPA) is actively seeking companies engaged in developing biohybrid robots—robots that combine synthetic and biological components. This initiative aims to push the boundaries of current robotics technology by integrating living tissues with mechanical systems, potentially leading to more adaptable, efficient, and versatile robotic platforms. DARPA’s funding call highlights its commitment to advancing robotics research beyond existing capabilities, encouraging innovation in the creation of hybrid systems that leverage the strengths of both biological and synthetic elements. This move could open new avenues in robotics applications, including enhanced mobility, self-repair, and responsiveness, although specific project details and timelines were not provided in the article.
robotbiohybrid-robotsDARPAsynthetic-biologyrobotics-researchadvanced-roboticsbio-roboticsWhy humanoid robots aren't advancing as fast as AI chatbots - The Robot Report
The article discusses why humanoid robots are not advancing as rapidly as AI chatbots, despite recent breakthroughs in large language models (LLMs) that power conversational AI. While tech leaders like Elon Musk and Jensen Huang predict humanoid robots will soon perform complex tasks such as surgery or home assistance, robotics experts like UC Berkeley's Ken Goldberg caution that these expectations are overly optimistic. Goldberg highlights a fundamental challenge known as the “100,000-year data gap,” referring to the vast difference between the extensive textual data available to train AI chatbots and the limited physical interaction data available to train robots for real-world tasks. This gap significantly slows the development of robots’ dexterity and manipulation skills, which remain far behind their language processing capabilities. Goldberg emphasizes that the core difficulty lies in robots’ ability to perform precise physical tasks, such as picking up a wine glass or changing a light bulb—actions humans do effortlessly but robots struggle with due to the complexity of spatial perception and fine motor control. This issue, known
roboticshumanoid-robotsAI-chatbotsmachine-learningautomationrobotics-researchartificial-intelligenceHumanoid robot uses human data to master cartwheels and sprints
Researchers at Cornell University have developed BeyondMimic, a novel framework enabling humanoid robots to perform complex, fluid human-like motions such as cartwheels, sprints, dance moves, and even Cristiano Ronaldo’s “Siu” celebration. Unlike traditional programming methods that require task-specific coding, BeyondMimic uses human motion capture data to train robots through a unified policy, allowing them to generalize and execute new tasks without prior training. This system leverages Markov Decision Processes and hyperparameters to seamlessly transition between diverse movements while preserving the style, timing, and expression of the original human actions. A key innovation in BeyondMimic is the use of loss-guided diffusion, which guides the robot’s real-time movements via differentiable cost functions, ensuring accuracy, flexibility, balance, and stability. The framework supports various real-world robotic controls such as path following, joystick operation, and obstacle avoidance, making it highly adaptable. The entire training pipeline is open-source and reproducible, providing a
roboticshumanoid-robotmotion-trackingmachine-learningrobot-controlartificial-intelligencerobotics-researchMIT roboticists debate the future of robotics, data, and computing - The Robot Report
At the IEEE International Conference on Robotics and Automation (ICRA), leading roboticists debated the future direction of robotics, focusing on whether advances will be driven primarily by code-based models or data-driven approaches. The panel, moderated by Ken Goldberg of UC Berkeley and featuring experts such as Daniela Rus, Russ Tedrake, Leslie Kaelbling, and others, highlighted a growing divide in the field. Rus and Tedrake strongly advocated for data-centric methods, emphasizing that real-world robotics requires machines to learn from extensive, multimodal datasets capturing human actions and environmental variability. They argued that traditional physics-based models work well in controlled settings but fail to generalize to unpredictable, human-centered tasks. Rus’s team at MIT’s CSAIL is pioneering this approach by collecting detailed sensor data on everyday human activities like cooking, capturing nuances such as gaze and force interactions to train AI systems that enable robots to generalize and adapt. Tedrake illustrated how scaling data enables robots to develop "common sense" for dexter
roboticsartificial-intelligencemachine-learningrobotics-researchdata-driven-roboticshuman-robot-interactionrobotic-automationSensing robot hand flicks, flinches, and grips like a human
A student team at USC Viterbi, led by assistant professor Daniel Seita, has developed the MOTIF Hand, a robotic hand designed to mimic human touch by sensing multiple modalities such as pressure, temperature, and motion. Unlike traditional robot grippers, the MOTIF Hand integrates a thermal camera embedded in its palm to detect heat without physical contact, allowing it to "flinch" away from hot surfaces much like a human would. It also uses force sensors in its fingers to apply precise pressure and can gauge the weight or contents of objects by flicking or shaking them, replicating human instincts in object interaction. The MOTIF Hand builds on previous open-source designs like Carnegie Mellon’s LEAP Hand, with the USC team also committing to open-source their work to foster collaboration in the robotics community. The developers emphasize that this platform is intended as a foundation for further research, aiming to make advanced tactile sensing accessible to more teams. Their findings have been published on Arxiv, highlighting a significant step toward
robotrobotic-handsensorshuman-robot-interactiontactile-sensingthermal-detectionrobotics-researchFlying squirrels' scaly tails inspire next-gen bionic robots, drones
Researchers at Empa in Switzerland, in collaboration with the Max Planck Institute in Germany, have studied the unique tail mechanics of African scaly-tailed squirrels to inspire the development of next-generation bionic robots and drones. These squirrels possess thorn-covered scales on the underside of their tails, which provide exceptional grip and stability when clinging to smooth or uneven tree bark. While biologists had long suspected the scales aid in climbing, this study is the first to scientifically test and confirm their role in preventing slipping and enhancing stability. The research team used a combination of analytical models and physical replicas, including 3D-printed artificial squirrels with scaled tails and claws, to experimentally validate how these tail spines contribute to secure perching. Moving forward, the team plans to incorporate dynamic movement into their models to better understand how the scaly tail absorbs impact and stabilizes the squirrels during rapid or emergency landings, such as when evading predators mid-glide. Ultimately, these insights into natural locomotion could inform
robotbionic-robotsdronesbiomimicryrobotics-researchsoft-roboticsenergy-efficient-robotics