Scaling General Physical AI: How RIVR is Unlocking the Next Era of Intelligent Robots

Technology
31
March
2025
The race to achieve General Physical AI for robots demands an immense volume of meaningful data—yet such data remains largely unavailable in robotics. At RIVR, we are pioneering General Physical AI by deploying robots in last-mile delivery, giving one human the power of 1,000 robots. Our approach fuses artificial neural networks with a breakthrough wheeled-leg robot design, enhancing efficiency, sustainability, and scalability in urban mobility. With a mission to deploy 1M+ robots, we are setting the foundation for the data flywheel that fuels robotic intelligence.
Article Summary
Author
RIVR
We give 1 human the power of a 1000.

As we gear up for NVIDIA GTC 2025, we’re excited to showcase how our deep learning methodologies are driving this transformation—advancing mobility, manipulation, and autonomous navigation at scale.

The Deep Learning Breakthrough

We integrate both Reinforcement Learning (RL) and Supervised Learning (SL) into our neural networks, developing a robust Physical AI that mimics human learning—combining imitation with trial-and-error learning. This integrated strategy allows us to handle robotic challenges through a unified neural network architecture, streamlining software coordination and reducing computational demands.

RL with Simulated Data: The Foundation of Rapid Learning

Our approach to RL is built on large-scale simulation. Using NVIDIA Isaac Sim, a reference robotic simulation application and NVIDIA Isaac Lab - an open source framework for robot learning, we train our legged robots through trial-and-error, dramatically reducing training time. This method allows us to:

  • Simulate complex real-world physics, including electric actuators and sensor models.
  • Develop proprietary learning algorithms that optimize robot behavior.
  • Create game-like training worlds that refine mobility, manipulation and autonomy.

By the time our robots are deployed, they have already encountered and adapted to millions of real-world scenarios in simulation, allowing them to master new tasks in just days.

SL with Real-World Data: The Data Flywheel

To complement RL, we leverage SL with real-world data, similar to techniques used in autonomous driving. By collecting expert data in the field, our robots continuously refine their understanding of complex urban environments. This data flywheel enables each deployed robot to gather more insights, enhancing autonomy and accelerating deployment. As more robots enter the field, they generate even richer data, ultimately solving the long tail of autonomy.

Why NVIDIA GTC Matters

As an early collaborator with NVIDIA since 2019 through ETH Zurich, RIVR’s founders have played a key role in shaping GPU-accelerated simulation technologies. At GTC 2025, we’re excited to share how our advancements in Physical AI are transforming last-mile delivery and paving the way for intelligent robotic systems at scale.

The future of robotics will be driven by those who can deploy at scale and unlock the data needed to train General Physical AI. At RIVR, we’re not just developing robots—we’re shaping an ecosystem where AI-powered robots redefine urban logistics.

We look forward to connecting with industry leaders, researchers, and AI pioneers at GTC to drive the next frontier of intelligent automation.

Related Stories