- Intrinsic, an Alphabet X spinout, incorporates Nvidia technologies into its Flowstate robotic app platform.
- Key integration includes Nvidia’s Isaac Manipulator for creating workflows for robot arms.
- Collaboration targets enhancing grasping capabilities crucial for manufacturing and fulfillment automation.
- Systems trained on extensive datasets aim for hardware-agnostic tasks across diverse objects.
- Intrinsic also collaborates with DeepMind on pose estimation and path planning.
- Testing involves Isaac Sim for simulation and real-world application trials.
- Advancements enable efficient object manipulation and seamless coordination of multiple robots.
- Intrinsic explores dual-arm systems, indicating future opportunities in humanoid robotics.
Main AI News:
Breaking news from the Automate conference: Intrinsic, a spinout from Alphabet X, announced a strategic move at the Chicago event on Monday. The company revealed its integration of several Nvidia technologies into its Flowstate robotic application platform.
The highlight of this collaboration is the incorporation of Nvidia’s Isaac Manipulator, a suite of foundational models tailored for creating workflows for robot arms. Launched at GTC in March, this offering has already gained traction among major players in industrial automation, such as Yaskawa, Solomon, PickNik Robotics, Ready Robotics, Franka Robotics, and Universal Robots.
The primary focus of this partnership is on grasping — a critical function for both manufacturing and fulfillment automation. These systems are trained on extensive datasets, aiming to perform tasks seamlessly across various hardware setups (i.e., hardware agnosticism) and with different types of objects.
The objective is to enable transferable picking methods across different scenarios, akin to human adaptability in handling objects. While humans can adjust their picking actions to different objects and settings, robots often lack this capability — at least for now.
“In the future, developers will leverage pre-built universal grasping skills like these to expedite their programming processes significantly,” remarked Wendy Tan White, Founder and CEO of Intrinsic. “This advancement underscores how foundational models could revolutionize the industry by simplifying robot programming challenges at scale, unlocking new applications, reducing development costs, and enhancing flexibility for end users.”
Initial testing of Flowstate took place in Isaac Sim — Nvidia’s robotic simulation platform. Trumpf Machine Tools, an Intrinsic customer, has been actively involved in testing a prototype of the system.
White highlighted Trumpf’s work with the platform, stating, “This universal grasping skill, trained with 100% synthetic data in Isaac Sim, can be utilized to develop sophisticated solutions capable of adaptive and versatile object grasping tasks in simulation and real-world environments.”
Moreover, Intrinsic is collaborating with DeepMind, another Alphabet-owned entity, to tackle pose estimation and path planning — two critical components of automation. For path planning, the system underwent training on a dataset comprising over 130,000 objects. Intrinsic claims that its systems can determine object orientations in a matter of seconds, a crucial capability for effective object manipulation.
Another significant aspect of Intrinsic’s collaboration with DeepMind is the orchestration of multiple robots simultaneously. White elaborated, “Our teams have successfully tested this 100% ML-generated solution to seamlessly coordinate four separate robots engaged in a scaled-down car welding application simulation.” She added, “The motion plans and trajectories for each robot are auto-generated, collision-free, and remarkably efficient — outperforming traditional methods by approximately 25%.”
Furthermore, Intrinsic is exploring systems that employ dual-arm configurations, aligning with the evolving landscape of humanoid robots. This trend is expected to gain momentum in the coming years, offering a myriad of new applications for such systems. Transitioning from single-arm to dual-arm setups holds promise for unlocking a host of additional capabilities in robotic technology.
Conclusion:
The integration of Nvidia technologies into Intrinsic’s robotic platform signifies a significant leap forward in automation capabilities. By leveraging advanced models and simulations, Intrinsic is poised to streamline programming processes, reduce development costs, and unlock new applications. This collaboration underscores the growing synergy between AI and robotics, paving the way for transformative advancements in various industries.
