Coco Robotics has introduced Coco 2, its new autonomous delivery robots designed to expand the company’s operations across urban environments. The platform moves from human-guided robotics to fully autonomous systems, according to the company. Coco 2 has been developed using data collected from millions of miles of urban operations.
The company said the robots are designed to adapt to new cities and navigate complex environments, including varying weather conditions and dense traffic. Operational data from deployments in cities such as Miami, Chicago, and Los Angeles has been used to train the system’s artificial intelligence models, with the stated aim of improving performance and enabling expansion into additional markets.
The company said Coco 2 is designed as a general-purpose urban robotics platform capable of transporting goods for restaurants, grocery stores, pharmacies, and other retailers. While earlier deployments focused primarily on sidewalk deliveries, the new model is intended to operate on sidewalks, bike lanes, and roads where regulations permit. Coco stated that the updated system can reduce delivery times by up to 50 percent compared with its previous generation and increase uptime and daily order capacity.
Zach Rash, co-founder and chief executive officer of Coco Robotics, said the company’s development process has relied on operational feedback from real-world deployments. “Every mile our robots have driven has made the whole fleet smarter,” Rash said. “Human-in-the-loop learnings have helped us improve with every edge case, creating a feedback loop between deployment, data collection, and model advancements. This ongoing process has steadily enhanced our fleet’s intelligence, enabling Coco to operate in new cities with real-time adaptability.”
Coco currently provides autonomous delivery services through platforms including Uber Eats, DoorDash, and Wolt, and reports serving more than 3,000 merchants and restaurants. The company said it plans to scale its fleet to thousands of robots globally by the end of the year.
The Coco 2 system incorporates simulation and computing technologies from NVIDIA. Using NVIDIA Omniverse libraries and the Isaac Sim and Isaac Lab frameworks, along with NVIDIA Cosmos models, the company generates simulated urban scenarios to train its artificial intelligence systems. The robots operate on NVIDIA Jetson Orin NX computing modules, which process data on-device rather than relying primarily on cloud connectivity.
Amit Goel, head of strategic partnerships at NVIDIA, said the collaboration supports the broader deployment of autonomous systems. “The era of physical AI has arrived, and scaling it requires a seamless loop between massive real-world data and high-performance edge computing,” Goel said. “By leveraging NVIDIA’s full robotics stack, Coco is accelerating the deployment of autonomous systems that can safely navigate the complexities of our cities at scale.”
