
NVIDIA Blog: Practice Makes Precise: At CoRL, NVIDIA and Partners Reveal How Robots Rehearse Delicate Work in Simulation
Simulation, which has long been the proving ground for robots learning to walk, roll and hop, is now helping researchers teach robots tasks that require a finer touch: aligning gears, connecting cables and assembling small components. The shift from locomotion to manipulation is taking center stage at this week’s Conference on Robot Learning (CoRL) in Austin, Texas, alongside the rise of agentic robotics — a new paradigm that integrates frontier agents to create interpretable code for physical
Robots are learning to perform delicate manipulation tasks like assembling components and connecting cables through simulation-based training, a shift from previous focus on locomotion. A world action model trained on diverse robot and human data can perform complex tasks with high precision and transfer skills between different robot types with minimal demonstrations. Simulation frameworks with accurate physics modeling allow researchers to train thousands of robots in parallel before deploying them on physical hardware, enabling the precision and adaptability needed for industrial deployment. This represents a move toward agentic robotics, where AI agents can autonomously sequence skills and recover from errors during task execution.
Want to go deeper than the news? Explore live, cohort-based AI courses taught by practitioners.
Browse AI courses on Maven
Bridging the divide between deterministic physics-based control and policy-driven adaptation, Boston Dynamics has deployed a hybrid locomotion architecture on a wheeled variant of its Atlas humanoid. The “experiment” combined classical model-based motor control with whole-body reinforcement learning to maintain dynamic stability across unstructured, real-world terrain. In the video, Andy Wu, controls engineer for Atlas at Boston Dynamics, highlights some recent experiments that he

Agility arrives on the Nasdaq today as the first publicly-traded pure-play humanoid robotics company.
Also reported by Ars Technica

The new Stretch features an updated modular design that enables “plug and play” part swaps. | Source: Boston Dynamics Boston Dynamics Inc. this week unveiled Stretch Gen 2, the latest generation of its autonomous mobile trailer unloading robot. With over 100,000 operating hours and tens of millions of boxes handled under its belt, the company built the updated robot to maximize uptime and streamline maintenance. Brian Donnelly, the vice president and general manager of logistics robotics at Bost

Physicl is trying to solve the sim-to-real gap with a pipeline originally built to market goods online.

Construction companies are struggling to find enough workers. For startups, that means venture investors are pouring money into companies betting that robots can help close the gap. Companies spanning construction and robotics have raised about $1.6 billion this year through Oct. 29 — more than 4x the $389 million recorded in all of 2025 — according to Crunchbase data. The surge was driven by three companies that raised the majority of the funding. The broad trend: For contractors struggling to

Wi-Fi connectivity needs to be reliable for real-time mobile robot management. Source: Synaptics An autonomous mobile robot or AMR may cross dozens of Wi-Fi coverage zones during a single shift. Each transition creates a small but potentially consequential communications problem: How does the robot find its next access point without interrupting the connection it is already using? The answer matters because AMRs, industrial inspection robots, warehouse automation, and healthcare service robots i