This paper presents a comprehensive review of diffusion models in robotics, focusing on their application in robot learning, data scaling, reinforcement learning, and imitation learning. The authors conducted a systematic literature review, selecting peer-reviewed articles and high-impact preprints to analyze the current state of diffusion model research in robotics. The paper discusses various applications of diffusion models, including semantic-level data augmentation, cross-view and morphology synthesis, automated simulation asset and task generation, trajectory optimization, policy representation, hierarchical planning, and real-time deployment. It also addresses the challenges and limitations of current approaches, such as computational cost, inference latency, and data efficiency, while highlighting promising future directions for research and development in the field.
This paper presents LeaP, a learnable source prior for generative robot policies, which replaces fixed standard Gaussian distributions with state-dependent random initialization. LeaP jointly learns the mean and variance of an initial Gaussian distribution, improving action generation in robotic tasks. The method achieves an average success rate of 81.6% in simulation and 80.0% on real robotic arms, outperforming existing baselines like A2A and VITA. The paper is accepted at CoRL 2026, and the code is open-sourced.
Skild AI presents a research paper detailing the development of an AI model trained across a vast array of robot bodies, enabling it to adapt to unpredictable scenarios without prior exposure. The model demonstrates zero-shot control and in-context learning, showing resilience in scenarios like limb loss, joint failure, and morphological changes. The work highlights the importance of adaptability in embodied AI for real-world applications.
Professor Shi Qing's team from Beijing Institute of Technology was inspired by the high-speed running mechanism of the elephant shrew. They designed a micro quadruped robot named FLEXOR equipped with a dual-joint coupled spine. Through dynamic spine-leg coordination, the robot achieved a 31.6% increase in speed and a 32.2% reduction in energy consumption without upgrading its motors. The study reveals the critical role of the timing coordination between the spine and legs in locomotion performance and verifies the universality of this mechanism across different robot morphologies, providing new insights for highly mobile embodied intelligent robots.
The article analyzes China's leading position in the field of humanoid robot patents, noting that as of the first half of 2026, China holds 60.8% of relevant patents, significantly surpassing the United States and South Korea. The article examines the evolution of patent strategies, the trend of technological shift from hardware to software, and the progress made by Chinese companies in patent quality, application scenarios, and global layout. It also points out that patent quantity does not equate to comprehensive technological leadership, emphasizing the challenges of manufacturing costs and quality, as well as the dual role of patents in industrial competition.
Researchers from North Carolina State University have developed a soft, non-electronic robot that jumps continuously using infrared light. The 'soft ring jumper' uses elastic energy stored through torsion, with a V-shaped aluminum stopper controlling the direction of movement. By adjusting the geometry of the V-stop, the robot can crawl, jump forward, or leap vertically. The design enables autonomous jumping without external resets, opening possibilities for use in environmental navigation and unstructured terrain.
Carnegie Mellon University researchers have developed LAMP, a system that allows multiple robots to work together to move objects through crowded spaces. LAMP combines learned models with search-based planning to enable efficient navigation and coordination, with successful testing in complex environments and a demonstration at the 2026 IROS conference.
Cornell researchers have developed microscopic robots that can sense temperature and work together to alter their environment by pumping liquid between hot and cold regions. Inspired by natural systems like fireflies and cilia, these robots use synchronized motion and electronic communication to achieve collective behavior, marking a significant step toward advanced micro-robotics with potential applications in medical and agricultural settings.
MIT researchers have developed a reconfigurable robotic lab that autonomously assembles, tunes, and dismantles optical experiments. The system uses robotic arms, 3D-printed component housings, and a cloud-based interface to enable remote operation. It can build and fine-tune a laser cavity in under 30 minutes, demonstrating the potential for fully automated optical experiments.
This research introduces a novel algorithm for generating grasp keypoints using straight skeletons to identify high-potential regions for stable grasping. A fully automated pipeline enables large-scale annotation without human intervention, and a new architecture, SkelGG-CNN, is introduced for 4-DoF grasping. The method is extended to 6-DoF grasping using a generalized quasi-3D skeletonization algorithm, achieving high accuracy on multiple benchmarks and real-world experiments. The framework enables efficient training of lightweight models for real-world deployment.
MIT researchers have developed a new technique that helps generative AI models meet strict safety and task-specific constraints without sacrificing output quality. The method, called HardFlow, reformulates constraint satisfaction as a trajectory-optim, allowing models to explore more freely during generation while ensuring final compliance with hard constraints. It is tested on robotics, control, and computer vision tasks, consistently outperforming existing methods in constraint satisfaction and solution quality.
This article explores how tactile data is helping robots improve their dexterous manipulation skills. Researchers are creating large tactile datasets and models that can use tactile feedback to enhance robot performance in tasks like folding laundry or turning keys. Challenges include the difficulty of integrating tactile data with vision-based models and the need for more diverse and scalable datasets to achieve significant improvements in robot dexterity.
Researchers at the University of Queensland have developed 'Paraborgs'—cyborg cockroaches equipped with needles and cameras to assist in search-and-rescue missions. These insects, grafted with electronic interfaces, can navigate rubble, deliver drugs, and operate in teams. The study addresses challenges in autonomy, communication, and human interaction with the technology.
Engineers at MIT and EPFL have developed a flapping-wing robot, or FAAV, that can swim underwater and then flap into the air, mimicking the movement of diving birds. The robot, weighing less than 300 grams, can transition between water and air using optimized wing size, flapping frequency, and tail angle. The study, published in Science, could enable new aerial-aquatic drones for oceanographic research and environmental monitoring.
MIT researchers have developed FloatForm, a system of small robotic boats that can autonomously assemble into floating structures, disassemble, and reconfigure. Inspired by ant rafts, the system uses decentralized control and modular components to create dynamic, programmable water-based infrastructure. The work has potential applications in urban planning, emergency response, and adaptive public spaces.
NASA's AstroPix gamma-ray sensor technology will be tested on a robotic arm demonstration mission as part of the Fly Foundational Robots mission, set to launch in late 2027. The sensors are designed to measure gamma rays between 20,000 and 700,000 electron volts, addressing gaps in current detection capabilities for high-energy gamma rays. The mission also includes a robotic servicing demonstration and will host the AstroPix payload in an Orbital Replacement Unit.
Sanctuary AI has demonstrated zero-shot in-hand manipulation using their proprietary hydraulic hand, autonomously reorienting a lettered cube ten times without dropping it. This showcases successful simulation-to-reality transfer, overcoming challenges in simulating dexterous hands and highlighting the company's advanced robotic hardware and policy training capabilities.
Physical AI refers to AI systems that enable robots to perform physical tasks in the real world. The article explains the immense difficulty of this endeavor, highlighting the variability of the physical environment, the need for multiple AI capabilities to work together, and the importance of safety and reliability. Dexterity's compositional approach to AI architecture is presented as a solution that enables production-scale Physical AI across multiple industries.
Dexterity's research introduces 'Transactable World Models' as a core component for Physical AI, treating world models as operators that reason about physical reality rather than storing data. These models enable robust manipulation by integrating physics, handling uncertainty, and supporting multi-agent coordination with explicit rollback and transaction guarantees. The approach emphasizes interpretability, real-time consistency, and the ability to reason about cause-and-effect relationships in dynamic environments.
