Food Delivery Robot Coming to Fremont Streets

The in-house developed DoorDash "Dot" robot is a product of DoorDash Labs and is manufactured locally in Fremont by Sonic Manufacturing. This localization is part of Fremont's effort to be a testbed for innovation while supporting local advanced manufacturers. The city has collaborated with DoorDash's economic development and public works departments to structure the robot's introduction to the community. Dot's autonomy stack relies on a sensor suite that includes lidar, cameras, and radar to navigate complex environments like parking lots and busy streets. Its perception system is primarily vision-based, utilizing eight external cameras for 360-degree coverage, supplemented by four radar units and three high-resolution lidar sensors for situational awareness. The navigation software combines deep learning with search-based algorithms to interpret the behavior of other road users and plot a safe and predictable path. The Fremont deployment is governed by a city-issued encroachment permit, which outlines a phased rollout and sets specific operational parameters. The initial phase will see three human-chaperoned robots mapping the service area, which includes Downtown, City Center, Centerville, and Irvington. This will be followed by a second phase with up to 30 robots operating under remote monitoring. Speed is capped at 5 mph on sidewalks, 16 mph in bike lanes, and 20 mph on neighborhood streets. Public demonstrations are scheduled to coincide with Fremont's Restaurant Week from March 6th to 15th, with the first delivery on March 5th. This controlled introduction is designed to build community trust and gather feedback. Concerns have already been raised by local community groups, particularly regarding the safety of cyclists and seniors who worry about the robots making sudden stops. The Dot robot is part of a larger DoorDash strategy to create a multi-modal delivery ecosystem. An AI-powered dispatch system, the Autonomous Delivery Platform, will decide whether an order is best suited for a human Dasher, a drone, or a robot like Dot based on factors like order size and distance. When encountering unexpected situations on the road, the robots are designed to stop and await assistance from a remote human operator. This Fremont launch follows a pilot program in the Phoenix metropolitan area, where the public and merchant reception was reported to be "overwhelmingly positive." The robot has undergone extensive testing over hundreds of thousands of miles on public roads, navigating through real-world challenges such as construction zones and emergency vehicles. While DoorDash has detailed its sensor and high-level software approach, specifics on the underlying processing hardware, such as whether it uses platforms like the NVIDIA Jetson Orin SoC common in other autonomous robots, have not been disclosed. The Jetson Orin platform provides significant AI performance, up to 275 TOPS, and is designed for complex robotics applications with its Ampere architecture GPU and deep learning accelerators. From a connectivity standpoint, the broader IoT landscape for devices like delivery robots is moving towards standards like Matter and Thread to ensure interoperability between different manufacturers' products. These protocols operate over IP, simplifying communication and control for a diverse range of smart devices, which could be relevant for future integrations and fleet management in complex urban environments.