Over the course of four years of extensive cooperation with Refraction AI, MobiDev provided full-cycle software development for the robotic delivery solution. After the service went live in early 2020, Refraction AI continued to grow and evolve, attracting a total investment of $10.7M.
Country
USA
Industry
Retail, Logistics and Transportation, Hospitality
Cooperation period
2021-2024
Cooperation type
Consulting, dedicated development team of 6 experts
Country
USA
Industry
Retail, Logistics and Transportation, Hospitality
Cooperation period
2021-2024
Cooperation type
Consulting, dedicated development team of 6 experts
When Refraction AI’s founders came out of stealth and presented the first generation of their self-driving delivery vehicle REV-1 during TechCrunch Sessions “Mobility” in the summer of 2019, they could not have foreseen how dramatically the world would change within six months. The COVID-19 pandemic created an urgent need for contactless delivery solutions, positioning robotic innovations like REV-1 as critical lifelines for hospitality businesses during lockdowns.
The global pandemic crisis not only accelerated the demand for autonomous delivery but also fostered greater openness among consumers. The success of its pilot period propelled Refraction AI’s growth, earning significant attention and attracting a total investment of $10.7M.
Recognizing their innovation and impact, Refraction AI was awarded the AI Breakthrough Award as the Best Robotics Sensing Solution in 2020.
In 2022, Refraction AI formed a remarkable partnership with Chick-fil-A, the largest chain specializing in chicken sandwiches. The collaboration aimed to integrate delivery robots to enhance restaurant operations, reduce delivery costs, and ensure food meets customer expectations. A year later, the partnership expanded when Chick-fil-A opened a new location in Alpharetta, Georgia, introducing delivery robots as part of its innovative services.
The idea for Refraction AI was conceived by two University of Michigan professors, Matthew Johnson-Roberson and Ram Vasudevan. With solid academic experience in robotics and their work at the University of Michigan’s Robotics Program, they sought to leverage robotics to lower the cost of last-mile delivery and create a sustainable approach for autonomous food delivery. Their solution aimed to navigate both vehicle and pedestrian traffic, as well as adverse weather conditions like rain and snow.
This vision materialized as REV-1, a delivery robot classified as an e-bike, which was first presented at TC Sessions “Mobility” in 2019. The team reduced the cost of each delivery robot from $10,000 to $5,000, in particular by using onboard cameras instead of expensive lidar sensors.
REV-1 entered the field in late 2019 and early 2020, when Refraction AI launched a pilot program for commercial food delivery in Ann Arbor, Michigan. This pilot served four local restaurants and a few hundred participants who signed up via Refraction AI’s website.
A few months later, the COVID-19 pandemic brought strict restrictions on visiting public spaces and mandated social distancing. Contactless delivery became essential for businesses to stay operational and for customers to receive food and goods safely.
In this rapidly changing environment, Refraction AI expanded its testing with local restaurant partners while searching for an outsourced team to support their in-house software development efforts. This marked the beginning of their collaboration with MobiDev, which initially focused on frontend development and later evolved into a full-cycle development, covering both frontend and backend activities.
One major issue in the food delivery sector is traffic congestion. In 2023, Americans lost an average of 42 hours annually to traffic congestion, costing each driver about $733 in lost time. Couriers often face challenges navigating crowded streets, especially during peak hours or in dense urban areas. A survey of delivery drivers for platforms like DoorDash, Grubhub, and Uber Eats revealed that just 7.5% of drivers consistently find curbside parking upon arrival, while 25% spend four to seven minutes searching for a parking spot. A study by the World Economic Forum predicts that by 2030, the number of delivery vehicles in the top 100 cities globally will increase by 36%, resulting in a 21% rise in traffic congestion.
Environmental concerns are also at the forefront. Traditional delivery methods relying on motor vehicles contribute to carbon emissions and pollution, particularly in cities striving to reduce their ecological footprint. A 2023 Statista study revealed that 60% of online consumers in Europe and North America are interested in carbon-neutral delivery options, reflecting growing environmental awareness.
COVID-19 introduced new challenges due to restrictions on in-person interactions. With health and safety protocols in place, minimizing physical contact became a priority. Traditional delivery methods became more challenging to manage and posed additional infection risks both for couriers and customers.
Before initiating development, the MobiDev team conducted tech strategy consulting to ensure smooth integration with Refraction AI’s existing system. This phase resulted in a roadmap for frontend development designed to align with Refraction AI’s business objectives, deliver the desired user experience, and utilize the relevant technologies within the agreed timelines. After Refraction AI expanded the MobiDev team to cover full-cycle development, we conducted a software audit of the existing backend.
Since Refraction AI already had in-house experts, it was crucial to establish an effective communication flow between them and the MobiDev team. While Refraction AI handled hardware setup and backend development, the MobiDev team focused on frontend development. Along with that, MobiDev excelled in creating comprehensive documentation for complex features and algorithms, which significantly streamlined the development process and improved team efficiency. Over time, Refraction AI expanded the collaboration, requesting MobiDev to cover full-cycle software development for further system upgrades.
Initially, the delivery robots could handle only one order at a time. However, Refraction AI created a new generation of robots equipped with compartments that could carry multiple orders in a single trip. The MobiDev team developed an automated order-assignment algorithm that optimized delivery routes by grouping orders traveling in the same direction while ensuring on-time delivery. This order batching feature allowed Refraction AI to manage complex conditions, such as prioritizing orders based on delivery times (e.g., immediate deliveries within a 30-minute window or scheduled orders for a specific time) and accounting for order cancellations. Additionally, the system was tailored to Refraction AI’s specific flow management, as the company worked with third-party systems like Bringg or OLO to receive orders, requiring real-time confirmation of whether a delivery could be completed within the required timeframe.
To ensure Refraction AI was optimized for handling increased order volumes while maintaining accurate and efficient delivery scheduling, the MobiDev team conducted rigorous performance and stress testing before launching the order batching functionality. These tests simulated peak loads, such as high volumes of incoming orders, to assess the system’s ability to process them without delays or errors. The testing phase also identified potential bottlenecks in the algorithm and ensured that real-time communication with third-party systems like Bringg and OLO remained seamless.
