AI in Logistics – Can Robot Minds Benefit The Supply Chain?
April 26, 2019
April 26, 2019
The idea of artificial intelligence isn’t new, having been around in various forms for several hundred years. The Mechanical Turk chess-playing automaton blew opponents away back in 1769 (though was later discovered to be an elaborate hoax involving a hidden chess master), the Turing Test was proposed in the 1950s as a means of testing machine intelligence, and nowadays AI is making its way deeper and deeper into our lives. But what of AI in logistics? As the supply chain becomes increasingly complex, it is more important than ever to weed out inefficiencies and streamline processes – and artificial intelligence can be a huge help throughout all stages of supply chain operations.
The primary differentiator between a true AI and a simple computer program are that AI can take in stochastic – unstructured and random – data, and make sense of it in a way that can lead to useful, intelligent outputs – a setup that the logistics industry is perfectly positioned to take advantage of. There is a similar distinction to be made between Robotic Process Automation (RPA), and true AI. RPA automates more basic tasks with a pre-determined set of rules, but can also be used to gather more data which can be fed to AI to improve its functioning. Both can be incredibly useful in logistics, but have quite different applications.
Predictive AI has the ability to plan ahead in ways that humans cannot. When it comes to route optimization for semi trucks, being able to take all of a route’s variables into account (traffic, weather, driver reliability, and more), and spit out an accurate idea of the best time to leave, roads to take, and which driver to give a certain load to, eliminates a lot of room for error from traditional setups.
So-called neural networks continually learn and improve their knowledge – in this example, of routes – almost in real-time, enabling them to teach themselves with minimal or no human input. In fact, this is realistically the only way for AI to better itself to the point where it is multitudes better than a human at the same job; a human simply can’t manually enter the quantities of data needed for an AI to process and evaluate routes with the necessary accuracy.
Last mile delivery
A McKinsey report found that the cost of last mile delivery – the final stretch a parcel travels to reach its recipient – nowadays constitutes upwards of 50% of the total cost of travel. This issue has a high equivalence with the “Travelling Salesman Problem” (TSP) or the “Vehicle Routing Problem” (VRP), which both boil down to what the quickest and most efficient route is for an individual or fleet, when having to target several spots in differing geographical areas and potentially at different times.
AI can help with this by analyzing large volumes of data and noticing any trends that may appear, no matter how unassuming they may seem to a human, and realizing the value in them. This can lead to highly optimized routes for fleets and individual drivers who can then make a much higher amount of deliveries or pickups in a shorter window of time, allowing cost and time savings to be made.
The concept of autonomous vehicles has been a frenzied media staple for quite a few years now. It can’t, however, be denied that the technology holds major promise. Tesla electric vehicles, for instance, have already clocked well over 1 billion miles on the road while using Tesla’s autopilot feature. The beauty of this lies in the fact that all Tesla cars are networked, and any data they take in via their cameras and sensors feed back to Tesla and allow its systems to learn and improve. You can imagine how over 1 billion miles of learning could lead to the self-driving skills shown in Tesla’s recently demonstrated video, as can be seen below.
When it comes to vehicles in the logistics sector, such as delivery trucks or full tractor trailers, plenty of issues remain in the way of large-scale deployment of autonomous technology. Autonomous trucks have a number of disadvantages compared to autonomous cars, in that they are simply far more dangerous vehicles on the road. A truck can’t quickly stop or nimbly change direction if an accident happens close ahead, and are also more difficult to outfit with the sensors necessary for full computer vision.
While we don’t doubt that this technology will make its way into trucks at some point in the future, drivers will remain indispensable for a long while to come, and even in the far future may remain in trucks to act more similarly to airline pilots than drivers. Technology such as truck platooning, emergency braking, lane assist, and overall electric tractor-trailers are features we expect to become more widespread long before complete automation.
With customers demanding ever-quicker shipping of their items, and e-commerce continuing to grow in demand, warehouses will have no choice but to become as optimized as humanly (or perhaps, robotically) possible.
With the help of AI, it will be possible to equip warehouses with a multitude of sensors to optimize pick rates, to notice when certain stocks are running low or are exhausted, and to be aware of and alert others to accidents (both potential and manifested). Startups such as Qopius and twentybn are already working on these solutions.
Besides simply observing what’s happening in the warehouse, robots will also begin to take more of an active part in the heavy lifting itself.
Boston Dynamics is pretty well known for their more flashy robotic dogs and humanoids, but last month they received a lot of attention for a video of their latest supply chain robots named “Handle” which makes the process of moving and stacking cartons look easy. They have now also acquired Kinema Systems, enabling their deep learning “Pick” technology to assist with the locating and moving of boxes on far more complex pallets than is currently possible. These types of robots will work hand-in-hand with others such as Autonomous Mobile Robots (AMRs), which can be loaded up and will traverse the warehouse to move parcels to wherever they’re needed next.
Even larger self-driving vehicles such as forklifts are already being deployed in warehouses, with an array of sensors making them generally safer than human drivers, and allowing them to communicate seamlessly with the rest of the AI-powered machines to obtain required inventory and plan the best route across the warehouse to deliver it.
It is likely that these robots will continue to work in conjunction with humans, who have the ability to think more dynamically and show more dexterity than any machine, while machines will take care of a lot more of the heavy lifting and riskier tasks required.
Back office A.I.
Back office operations are frequently ignored when it comes to AI in logistics. However, together with both RPA (the more basic, rule-based robotics) and more advanced AI, humans could enhance the quality and enjoyment of their work by eliminating the extremely repetitive tasks from their day-to-day and allowing them to focus on the more interpretive and “human” aspects of their work. This goes for many types of back office jobs in the logistics, supply chain, and trucking industries. Accountants, human resources, and data analysts are just a few of the positions which could greatly benefit from this robotic assistance.
An area as complex as logistics will find many uses for artificial intelligence. At the scale of national and global supply chains, optimizations of just a few percent can lead to enormous savings, and AI looks poised to be the breakthrough technology that can make this happen. The timeline of these implementations may still be some years off before being fully realized, but we’re no doubt already at the brink of a large push towards embracing AI benefits in the industry.
At FreightTrust, we’re using predictive AI to make route planning simpler and more efficient, allowing loads to be delivered on time and with greater reliability. If you’d like to find out more about what else our platforms can offer to freight brokers, carriers, and shippers, check out our product tour.
What We Do
Sign up now and save yourself time, money, and effort that you can reinvest elsewhere.