admin 
Autonomous delivery robots are beginning to move beyond pavements and into urban streets, as robotics company Coco Robotics deploys a faster generation of machines guided by mapping technology originally developed for the global mobile game Pokémon GO.
The company says its latest machine, known as Coco 2, can travel at speeds of up to 13 miles per hour while navigating city roads and bicycle lanes where local regulations permit. The system relies on spatial mapping and visual positioning tools from Niantic Spatial, a geospatial computing company that emerged from Niantic, the developer behind the augmented-reality game that attracted hundreds of millions of players worldwide.
Coco Robotics has built its business around small, wheeled robots designed to carry food and goods across urban neighbourhoods. More than half a million deliveries have already been completed by its fleet, which operates across cities in the United States and Europe. The firm runs roughly a thousand robots and serves thousands of merchants, including restaurants, grocery outlets and pharmacies that rely on quick local deliveries.
The introduction of Coco 2 marks an attempt to accelerate the evolution of “last-mile” delivery robots, shifting them from slower pavement-based systems to vehicles capable of travelling through a wider range of urban infrastructure. Earlier generations largely operated on sidewalks, moving at pedestrian speeds and often requiring remote human oversight. The new model is designed to operate autonomously and to adapt to streets, loading zones and cycle lanes while handling several orders at once.
Central to the strategy is Niantic Spatial’s Visual Positioning System, a technology that uses computer vision and geospatial artificial intelligence to determine a device’s exact location. The system compares camera images captured by a robot with a large database of mapped environments, allowing the machine to pinpoint its position even in areas where satellite signals are unreliable.
Urban navigation presents particular challenges for robots because tall buildings can interfere with GPS signals, creating what engineers describe as “urban canyons”. Delivery robots must also identify the correct entrance to buildings, avoid pedestrians and cyclists, and stop precisely at kerbs or loading areas. According to Niantic Spatial’s leadership, solving these challenges requires highly accurate localisation and detailed mapping of city environments.
Niantic’s mapping infrastructure draws on a vast archive of real-world images collected through augmented-reality applications. Players using smartphone cameras to capture landmarks and streets generated a database that now spans billions of visual data points across cities worldwide. Engineers have begun repurposing that information to train AI systems capable of interpreting the physical world in detail.
Coco Robotics says its robots are designed to learn from large amounts of operational data gathered during everyday deliveries. Millions of miles travelled in cities such as Los Angeles, Miami and Chicago have been used to refine the machine-learning models that guide the fleet. The robots have encountered a wide variety of real-world conditions, including heavy traffic, flooding and winter snow, allowing engineers to train algorithms to respond safely to unpredictable situations.
The next-generation model is also equipped with advanced edge-computing hardware capable of processing sensor data directly on the robot. This allows the machines to analyse camera feeds, detect obstacles and plan routes without relying heavily on cloud connections. Robotics developers argue that such onboard processing is essential for safe autonomy in crowded urban environments.
Expanding beyond pavements could reshape the economics of last-mile delivery. Coco Robotics says the ability to travel through bike lanes and roads shortens delivery routes and reduces travel times compared with earlier designs. The company claims the shift could lower delivery costs and increase the number of orders each robot can complete during a single operating cycle.
Competition in autonomous delivery has intensified as logistics firms search for ways to manage rising demand for fast deliveries and persistent labour shortages in urban courier services. Technology companies, robotics start-ups and e-commerce platforms have all explored robotic delivery vehicles, ranging from small ground robots to aerial drones.
Many attempts have struggled to scale because of safety concerns, regulatory uncertainty and technical limitations. Cities have imposed strict rules governing where robots may operate, and incidents involving delivery machines occasionally spark debate over pedestrian safety and liability.
Advocates of the technology argue that autonomous delivery could ease congestion and reduce emissions if robots replace short car journeys made by human couriers. Critics counter that widespread deployment may create new challenges, including cluttered pavements, cybersecurity risks and complex legal questions about responsibility when machines malfunction.
For Niantic Spatial, the collaboration signals a strategic shift from entertainment toward infrastructure for real-world computing. The company has increasingly positioned its geospatial technology as a foundation for applications beyond gaming, including robotics, urban planning and navigation systems.
Executives involved in the partnership describe the effort as part of a broader push to bring “spatial intelligence” into everyday devices and machines. By combining detailed mapping with artificial intelligence, they argue, robots can gain a deeper understanding of their surroundings and operate more reliably within complex city landscapes.
