Khalifa University’s Emirates ICT Innovation Center (EBTC) Develops Intuitu, an Advanced AI Tool to Drive Optimized Warehouse Deployment
The warehouses that pick, pack and dispatch our products and packages are getting smarter, more efficient, smaller, and even portable, thanks to recent advancements in digitalization, big data, and artificial intelligence (AI). As these ‘warehouses of the future’ shift from massive brick-and-mortar buildings to mobile containers closer to the customer, they present unique optimization challenges for the companies that use them.
This month, at the AI and Machine Learning Awards 2019 in London – a new line up of awards launched this year by Computing, UK’s leading business technology information resource – researchers from Khalifa University’s Emirates ICT Innovation Center (EBTIC) won two awards, including ‘Outstanding AI/Machine Learning Project’ and ‘Most Innovative use of AI/Machine Learning’ for their new cutting-edge tool for solving optimization problems related to warehouse deployment, called Intuitu.
Developed by researchers at EBTIC – an ICT research and innovation center founded by Khalifa University, Etisalat and BT, and supported by ICT Fund – and led by Dr. Sid Shakya and Dr. Rom Lee, together with BT, UK, Intuitu is an AI-based tool that optimizes the deployment of warehouses for large service organizations such as telecommunications companies. EBTIC is also working closely with its local partner Etisalat, one of the world’s leading telecom groups, to deploy the Intuitu tool to improve operational efficiency and faster service delivery of Etisalat’s mobile warehouses. They are also collaborating to leverage Intuitu’s advanced AI logic to manage Etisalat’s inventories and warehouses more efficiently.
“Recent advancements in internet-of-things (IoT) and connected technologies have greatly impacted warehouse design. They are getting smaller and mobile in their nature, shifting from fixed structures to mobile containers or lockers. It is a relatively new concept and poses challenges as redeployment has to happen frequently and optimally. Placing warehouses at the right place and at the right time can reduce travel and maximize productivity for field engineers,” said Dr. Sid Shakya, a Chief Researcher at EBTIC.
IoT enabled warehouses can be monitored and operated remotely. A fixed number of engineers and technicians are assigned to the warehouses based on their home location and working location. Ideally, the distance the workers have to travel to get to the warehouse and then to the client should be minimized. These mobile storages can be quickly deployed to different locations in a very short time. More importantly, they can be moved from one location to another and can be redeployed and reused, contributing to a more sustainable and eco-friendly model of warehousing.
One of the main goals of Intuitu is to determine the most ideal configuration of portable warehouses so that they are closer to consumers for faster delivery, while remaining close to technicians for optimal productivity. The tool drastically reduces the amount of time it would take for a technician to be deployed to provide a customer a requested service, which in turn will lead to significant efficiency gains.
Intuitu models the warehouse deployment problem as a combinatorial optimization problem – when mathematical techniques are applied to find optimal solutions within a finite set of possible solutions – and uses a sophisticated type of evolutionary algorithm known as genetic algorithm (GA) to find the optimal deployment design.
In this model, five key objectives were optimized: Minimizing the cumulative travel time for technicians to get spare parts; minimizing the distance from a warehouse to the served exchange sites; minimizing the distance from a warehouse to the home locations of the assigned technicians; minimizing the number of technicians that a warehouse serves; and lastly, minimizing the number of tasks that a warehouse is expected to serve per day.
Dr. Rom from the Inutitu team elaborated on how the tool works, “The genetic algorithm starts with a population of random solutions, which in this case, are a set of random warehouse locations, and performs a distance based clustering algorithm to find the sites that will be served by each location. It then calculates all five objectives and combines them together to evaluate the effectiveness of each solution. It then iteratively ‘evolves’ the population by applying genetic operators known as selection, crossover and mutation over a fixed number of allowed iterations, also known as generations, and consequently presents the best solution in the final population as the output.”
“Evolutionary algorithms and genetic algorithms are active areas of research in the AI community and EBTIC is at forefront of this research field,” Dr. Rom added.
“To make AI models operational, you must bring together a number of components including data, problem modelling, and identification of the right AI techniques. EBTIC approached this task in an agile fashion working closely with BT’s Applied Research team, and identified the right AI techniques to model the problem,” said Dr. Sid
Dr. Nawaf Al Moosa, Director of EBTIC, said: “EBTIC has a history of innovating practical AI solutions and have been leading in intellectual property generation in UAE for the past 10 years. These awards are a further testament to EBTIC’s successful history of applying AI techniques to the operations of its partner organizations, and to their effort to promote AI research in the UAE and the region.”
EBTIC is a foundational partner of Khalifa University’s new AI and Intelligent Systems Institute (AI Institute), a multidisciplinary research unit focused primarily on robotics, artificial intelligence, data science, next-gen networks, semiconductor technologies and cybersecurity. EBTIC and the AI Institute are committed to developing the key technologies required to bring the UAE significantly closer to reaching its goal of becoming a global hub for Artificial Intelligence innovation.
5 August 2019