Coco

Onboard software for sidewalk delivery robots and the road to autonomy

Watch the video

Coco runs remotely piloted sidewalk robots for last-mile food delivery on the sidewalks of Santa Monica, Los Angeles and Helsinki, Finland. Since around 2021, an xBerry team of robotics engineers has developed and maintained the robots’ onboard software stack. The result is a more reliable, better-performing fleet and a foundation for autonomous, pilot-less delivery.

Scroll pageg icon

Challenge

Food delivery demand grew fast, but the model still depended on human couriers. During the COVID-19 pandemic, this became a bottleneck: demand peaked while couriers were scarce and customers wanted contactless delivery.

Coco saw a better model – robots that deliver food reliably, at lower cost and with lower emissions. To scale that model to cities around the world, Coco needed two things:

  • a robust, high-performing onboard software stack for its robot fleet,

  • a long-term engineering partner with deep robotics and embedded systems experience.

Goals

A technical team built around Coco's product

Coco came to us with a working delivery robot and a clear business vision. The goal was to improve the product and its infrastructure so the fleet could scale to new cities.

We work with Coco in a team augmentation model. xBerry provides a complete team of total 6 robotics and embedded software engineers. Most of the engineering team has worked on Coco since the project started at xBerry. They work directly with Coco’s business leaders and product owners to:

  • identify emerging challenges,
  • define them as technical problems,
  • design and ship the right solutions.

Coco gets engineers who know its robots, codebase, and operations in depth without the cost of rebuilding that knowledge in-house.

Solution

  • We started with workshops with Coco to understand their needs and plan the cooperation.

  • We assembled 6 experts matched to Coco’s product vision and technical roadmap.

  • We develop the robots’ onboard software stack – ROS and C++ based software running on embedded Linux built with Yocto.

  • The work continues: the team is now developing autonomous driving capabilities for Coco’s robots.

About Coco

Coco is a robotic delivery service that uses remotely piloted sidewalk vehicles for last-mile delivery. Its goal is to make deliveries more affordable, reliable and sustainable for merchants and their customers.

Each robot is supervised by a trained operator - a "Coconaut" - and enters the public right-of-way only during an active delivery. The vehicles are shaped to move efficiently through dense urban areas. In Santa Monica, Los Angeles, Coco reports a 97% on-time delivery rate and a CO2 reduction of 0.25 kg per km compared with traditional courier delivery.

Results

With xBerry’s support, Coco has significantly improved the reliability and performance of its onboard software stack.

  • Higher operating efficiency for Coco’s fleet.
  • A better experience for Coco’s merchant partners and end customers.
  • A solid foundation for autonomy – Coco and xBerry are now developing autonomous driving capabilities for future pilot-less delivery trips.

Tech Stack

Team augmentation
ROS
C++
Linux embedded
Yocto
CUDA
Tensorflow
Python

What are the customers saying?

Their technical expertise and flexibility make xBerry an invaluable partner for us. It is crucial in the ever-changing and fast-paced environment to work with people that understand you and your needs.

Pratik Agarwal Head of AI at Coco

FAQ

Can delivery robots operate reliably on city sidewalks? Faq icon

Yes and Coco does so every day in Santa Monica, achieving a 97% on-time delivery rate. Sidewalks are an environment that can't be standardized, so reliability depends above all on the software running on board - which is the area xBerry works on with Coco.

How do you move from remotely piloted robots to autonomous delivery? Faq icon

In stages. Coco first built a reliable onboard software foundation for remotely piloted robots, and only then moved on to developing autonomous driving capabilities for pilot-less trips. Autonomy is built on top of a system that already works reliably, not instead of it.

What does team augmentation look like for a robotics company? Faq icon

In Coco's case, a complete team of xBerry engineers works with Coco's business leaders and product owners: understanding challenges, defining technical problems, and designing and implementing solutions within Coco's own product and roadmap.

What technologies are used in Coco's onboard software? Faq icon

ROS and C++ on embedded Linux built with Yocto, with CUDA and TensorFlow for machine learning, and Python.

Who can provide a robotics software team for autonomous robots? Faq icon

xBerry, a Physical AI R&D team based in Poland, works as an external engineering team for robotics companies - on onboard software, embedded Linux, ROS and autonomous capabilities. Coco, a robotic delivery service operating in Los Angeles, has worked with xBerry on its onboard stack and autonomy since 2021.

Planning a digital project?

Contact us Arrow icon