General Physical AI for Front-Door Delivery Robots

Title

General Physical AI for Front-Door Delivery Robots

Content

RIVR is pioneering General Physical AI by deploying robots in last-mile delivery, giving one human the power of 1,000 robots. Their approach fuses artificial neural networks with a breakthrough wheeled-leg design, enabling robots to move fluidly through complex urban terrain. With a mission to deploy over a million robots, RIVR is building a data flywheel that accelerates robotic intelligence with every delivery made. RIVR believes the answer lies in delivery, specifically last-mile and last-100-yard delivery. With over 166 billion packages delivered annually around the globe, this is a domain rich in general robotic challenges: mobility, manipulation, human interaction, and unstructured environments. Each delivery is a training repetition—a chance for the robot to navigate new terrain, interpret visual scenes, avoid pedestrians, open gates, and interact with humans. RIVR integrates both Reinforcement Learning (RL) and Supervised Learning (SL) into neural networks. Their RL approach uses large-scale simulation with parallel GPU simulators like NVIDIA Isaac Sim, training legged robots through trial-and-error learning. By the time robots are deployed, they have already encountered and adapted to millions of real-world scenarios in simulation. RIVR's partnership with Veho deploys stair-climbing delivery robots to ferry packages from vans directly to customers' front doors in Austin, Texas.

URL