Innovation is part of our DNA and with our investment to date, and our continuing research into areas such as artificial intelligence, robotics, automation and big data, we have an opportunity to create a sustainable technological advantage in an increasingly competitive market. With everything from websites to warehouses designed in-house, we have developed capabilities across a wide range of technologies.
Areas of Development
Automation and Robotics
Warehouse automation within our centralised CFCs is one of the key differentiators for Ocado in the online grocery retail market. Our most ambitious and innovative project of this sort has been the design and development of our proprietary ‘hive’ fulfilment solution, currently live in our Andover CFC, which will involve thousands of robots working together to retrieve from storage the groceries comprising a customer order.
In terms of picking robotics, we have two ambitious Horizon 2020 Research and Innovation projects called SecondHands and SoMa which combine state-of-the-art robotics, artificial intelligence, machine learning and advanced sensors to understand and assist human warehouse workers in real time.
For more information on this see the Development in Robotics Case Study
Data Science and Artificial Intelligence
Machine learning is a core competency at Ocado. Artificial intelligence is beginning to transform how customers and companies interact with each other. Ocado uses machine learning to develop predictive analytics, implement advanced monitoring and oversight, manage our operations, and achieve real-time optimisation of services. Our data science teams work on advanced projects using programming languages, tools and frameworks in areas related to computer vision, natural language processing and demand forecasting.
Big Data and the Cloud
We use advanced data analytics and cloud storage to deliver groceries to more than 645,000 active customers. Our cloud and data teams collect and process data from customer-facing websites, warehouses, delivery vans and beyond to provide a more personalised shopping experience to customers, optimise warehouse operations, empower the supply chain, speed decision-making and reduce costs.
Web and App Development
Ocado was the first grocery supermarket to launch Android, iOS and watchOS apps. With the move into a digital era, the retail sector is becoming increasingly diverse and requires the strategic use of applications to offer new immersive and interactive methods of shopping. Combining attractive aesthetics with pioneering functionality that offers unique ways to shop, we have helped to create applications which deliver an exciting, informative and tailored experience – while using intelligent technologies to remove the complexity of keeping it all managed and up to date.
Algorithms and Smart Optimisation
Our systems can make around four million routing calculations per second. Ocado's systems continuously monitor stock at the individual product level and compute multiple sales forecasts for every product in every warehouse. Warehouse replenishment systems are designed to automatically reorder stock from suppliers within the constraints of maximising product life and availability to customers while minimising waste and stock-holding.
Powering Ocado is a complex computing estate that covers everything from our webshop to our highly automated warehouses. Our distribution systems need to support the business as it evolves and changes, but at the same time balance workloads, performance and resources in an efficient manner. Our technology teams design and build these systems to be highly available and self-healing.
Simulation modelling is used at Ocado for various warehouse operations. Making highly optimal choices for physical devices, layout and algorithms is often beyond human capability, but getting it right is extremely important for any business. Companies can optimise by trial and error in production (often this is expensive, risky and slow) or they can create software to identify highly optimal choices. Ocado uses discrete event simulation for this purpose. This gives us deterministic, faster than real-time scenario and soak testing, allowing us to create systems that would otherwise be too complex to test and debug.
our proprietary routing algorithms
As we continue to grow our customer base, the routing algorithms that underpin the efficiency of our last mile delivery become increasingly complex. The algorithms must optimise the variants of thousands of delivery drivers in different cities with numerous drop destinations, with varying distances between them and calculate the shortest possible route for each delivery driver to complete their shift and end up back where they started. As we continue to grow, the scale of this problem keeps expanding.
Since 2007, when we outgrew the capacity of third party routing software, we have been using our internally developed proprietary algorithms, which employ real-time optimisation to help us decide which deliveries to assign to each van and the order in which they should be completed, while also ensuring we do our best to arrive on time for all our deliveries in time slots selected by the customer.
The algorithm makes several million route calculations per second to identify the best delivery routes for our drivers. We have continued to evolve our software over time so that we can assess which scenarios are most efficient.
Our platform is constantly running multiple instances of the optimiser simultaneously, each iteration focusing on a specific area for a certain day. This means that a customer can book a delivery slot online late on a Thursday night for delivery the following day, and the software can make sure it incorporates this new slot into the route optimisation for Friday. The software is quick and efficient, meaning that additions and revisions of delivery bookings can be taken into account when calculating the best possible solution within the time available. When you visit Ocado.com, the webshop is actually communicating directly with our routing software to establish which delivery slots are still available. When you click on the day you wish to receive your delivery in the calendar, the real time optimiser will return the available slots within 500 milliseconds!
The secret ingredient to our routing success is the broad range of variables we take into account when calculating the cost function, including van capacities, weights, volumes, fuel consumption and even driver experience. The work of our technology team means we regularly find ourselves in a position to solve our own problems without outsourcing to third parties. The close collaboration between our retail and technology teams continues to keep us a step ahead.
Developments in robotics
As part of our ongoing aim to lower costs and drive efficiencies within our model we have been evaluating the feasibility of robotic picking and packing within our CFCs through various initiatives, one of these being the SoMa project, a Horizon 2020 framework programme for research and innovation funded by the European Union.
One of the main challenges of robotic manipulation has been the handling of easily damageable and unpredictably shaped objects such as fruit and vegetables. To avoid damaging sensitive items, the project uses a compliant gripper (i.e. one that possesses spring-like properties) in conjunction with an industrial robot arm.
The variation in shape of the target objects imposes another set of constraints on the design of a suitable gripper. The gripper must be sufficiently versatile to pick a wide variety of products, including Ocado's current range which includes over 49,000 hypermarket items. The goal is to develop versatile, robust, cost-effective, and safe robotic grasping and manipulation capabilities.
An example of a compliant gripper is the RBO Hand 2. The gripper uses flexible rubber materials and pressurized air for passively adapting grasps which allows for safe and damage-free picking of objects. Our robotics team replicated a production warehouse scenario in order to evaluate the performance of the RBO Hand 2 for Ocado's use case. The team mounted the soft hand on two different robot arms.
We designed a set of experiments to evaluate grasping performance on an example set of artificial fruit stored in a tray. The experiments started with the simple scenario of grasping a single object. Results illustrated that the hand is able to successfully grasp a variety of shapes and provided us with insights on how to improve accuracy and success of picking going forward. Over the year we have continued to explore more complex scenarios, adding more objects and introducing additional environmental constraints that could be exploited by a grasping strategy.
More recently we have developed and trialled an alternative type of picking station, called OSPick. OSPick is an industrial robot designed to pick a range of groceries within our CFCs using suction. The system is straightforward in concept compared to many other industrial manipulation projects, but alongside this simplicity comes flexibility. The robotic pick station consists of a suction cup on the end of an articulated arm. The arm is equipped with a pipe running to an air compressor, which is capable of lifting items regardless of their deformability and shape. The biggest factors that influence the success of a pick are the item's weight and surface properties.
The future development and roll-out of this picking solution within our CFCs should allow us to reduce the cost of fulfilment within our operations and that of our commercial partners.