AMD acquired World Labs for $8.2 billion to gain its AI spatial intelligence capabilities. World Labs developed leading models for images, video, and spatial reconstruction. They recently released Atlas, a new omni model architecture that solves the sparse reconstruction problem through camera view prediction. Atlas has applications across robotics simulation, design, and real-world reconstruction. This acquisition significantly enhances AMD's position in spatial AI and related fields.
Charlie Kemp, cofounder and CTO of Hello Robot, develops mobile manipulators to assist people in homes and workplaces. His work focuses on creating assistive robots for older adults and individuals with disabilities, emphasizing close physical interaction. Kemp finds the most rewarding aspects of robotics in seeing software become physical reality and influencing the future. He advises aspiring roboticists to embrace real-world discovery and collaborate with non-robotics experts for impactful applications. His work has involved developing systems that enable meaningful physical assistance for people with mobility impairments.
Delta Intelligence's initial demonstration focused on household tasks, but founder Xiaojian Ma prioritizes energy and infrastructure as near-term deployment targets. The company is developing a head-and-backpack system to integrate models onto various humanoid platforms. Whole-body data collection was planned from the start, emphasizing the need to capture coordination beyond just the hands. Future plans involve scaling pretraining data, improving reinforcement learning, and conducting outdoor testing in harsh environments. The ultimate goal is to prove the demonstrated coordination can translate into reliable work in energy and infrastructure settings.
Dyna Robotics has teased a new robot, announcing that a fuller look is "coming soon." The teaser positions the forthcoming machine as being built for demanding, detailed work. This announcement follows Dyna's existing commercial deployments and work on its Dyna-2 model. The reveal is expected to provide further details on the robot's capabilities and intended applications. Further information regarding specifications and launch details is still forthcoming.
Figure joined NVIDIA’s Open Agent Safety Platform to establish boundaries for increasingly capable AI agents. This initiative combines the OpenShell runtime with Sentry, a monitoring and enforcement layer. Figure's participation focuses on ensuring the safety and trust of humanoid robots operating in homes and workplaces. The platform aims to constrain agent actions through runtime controls rather than just model safeguards. This collaboration provides a foundation for trustworthy agent systems where software decisions translate into physical actions.
MIT engineers developed a soft, muscle-powered swimming robot using a thin film of gel and genetically engineered muscle cells. These cells are designed to twitch in response to light, causing the robot's fins to flap and propel it through water. By flashing light on the fins, the team controlled the robot's direction and speed in a watery maze. The research demonstrated that this thin, two-dimensional design is a foundation for more efficient biohybrid robots. The study explored optimizing the gel's properties and groove shapes to maximize muscle force production.