If you follow tech news, you may have seen headlines about Amazon preparing to test humanoid robots for package delivery. The company’s latest initiative, based in a San Francisco office with a facility known as the “humanoid park,” is drawing attention for its futuristic vision: robots navigating obstacles, climbing stairs, and attempting to deliver packages.
Behind the eye-catching demos, Amazon’s humanoid delivery robot project is a complex and ambitious endeavour that is pushing the boundaries of what is possible in robotics.
The “Humanoid Park”: What do we know?
Amazon’s “humanoid park” is a tightly controlled indoor testing facility designed to simulate real-world delivery challenges. It includes mock doorways, steps, and a parked Rivian electric delivery van. The goal is to see how robots handle the unpredictable environments typical of American neighbourhoods.
Amazon is testing the Unitree G1 robot, manufactured by the Chinese robotics company Unitree. The G1 costs around $16,000 and features 23 joint motors and force-sensitive grippers. While the robot can climb stairs and carry packages, it is still in the early stages of development. In fact, it is being evaluated for its ability to perform basic delivery tasks.
This isn’t Amazon’s first venture into delivery robotics. In 2019, it launched Scout, a six-wheeled sidewalk delivery robot. Scout was quietly shelved in 2022 after public mishaps, such as freezing at the sight of squirrels and failing to recognise porch steps. Lessons from Scout’s deployment inform Amazon’s current approach to humanoid robots.
Hardware showdown: G1 vs. Digit
Amazon is pursuing a two-pronged robotics strategy. While the Unitree G1 is being tested for customer-facing “last mile” deliveries, Amazon is also piloting Agility Robotics’ Digit in warehouses. Digit is designed for stability and repetitive tasks, such as tote recycling, and is not intended for outdoor or last-mile delivery scenarios. Both robots are in early trial phases, and Amazon is using these pilots to evaluate their effectiveness in different settings.
Amazon is investing heavily in artificial intelligence to power its robotics programs. The company has created a new Agentic AI group focused on developing advanced AI systems for robotics. These systems aim to enable robots to perform complex, multistep tasks and interpret natural language commands. This software is central to improving robots’ ability to navigate unpredictable environments and safely interact with objects and people.
Amazon’s vision includes integrating humanoid robots, such as Rivian electric vans, with its delivery fleet. The plan is for robots to eventually exit the vans and deliver packages to customers’ doors, but for now, testing is limited to controlled environments and pilot programs.
Amazon has stated that robots are intended to complement, not replace, its human workforce. The company is hiring for roles related to robot maintenance and oversight. However, widespread deployment of delivery robots faces regulatory challenges, as only a few U.S. states currently allow autonomous sidewalk robots, and public acceptance varies.
The roadblocks ahead for Amazon
Amazon’s humanoid delivery robots remain in the early stages of development and are not yet ready for large-scale public deployment. The project is part of Amazon’s broader strategy to use AI and automation to improve logistics and delivery speed, but significant technical and regulatory challenges remain.
While Amazon refines its approach, competitors are also advancing. Tesla’s Optimus robot handles material transport at Gigafactories. Boston Dynamics’ Spot can open doors and climb rubble (though it’s not designed for deliveries), and ANYmal’s quadruped robots can navigate stairs and doorsteps autonomously. Some robotics experts suggest that wheeled or dog-like bots may prove more reliable and cost-effective than humanoid designs.
Amazon’s humanoid delivery robot initiative is a promising step towards the future of technology. While the technology is still experimental, the robots are being tested in a controlled facility, and the company’s investment in hardware and advanced AI is a beacon of hope for the challenges of real-world delivery.
