Simacan's improved estimated travel times (ETAs) for trucks:
Fast progress for slow movers
What if you know exactly at what time your truck or home delivery van arrives at their destinations? Not only for its next destination, but for the whole day? It will give you control of your operations. This is what the Simacan platform calculates for you. But how do we do that? And is a 100% score on accuracy of the Estimated Times of Arrival (ETAs) possible? Let’s dive into the magic of calculating ETAs on the Simacan Platform.
When planning or keeping track of your logistic operation, knowing how much time it takes for a vehicle to move to its destination is essential. Without this information, making a planning which closely matches reality could be considered hard, or even impossible. However, a planner can't know for sure how long it will take for a vehicle to move from point A to point B. After all, they simply can't just take a peek into the future.
Accurate ETAs
That’s why planners use Estimated Times of Arrival. It is important that these ETAs are as accurate as possible, such that the planner can make a planning which is as close to reality as possible. This offloads work during live operation. The more a planning matches reality, the less readjusting is required during live operation. Then you can truly focus on managing by exception: allowing the planning and support team to focus on the safety of the driver and goods, allowing them to come on time at their destination.
Instead of peeking into the future, we can look at the past and see how long it took other vehicles to move from point A to point B and use this information for calculating an ETA. Of course, capturing and storing every trip from every point A to every point B is infeasible, and we probably would still not have enough data points to make proper predictions. What about new or changed roads for instance? What if we have no records of vehicles driving over a specific road? It simply doesn't scale.
Instead, we use a map database, which is a dataset of all roads in the existing road network including their properties. It contains properties like, how fast are we allowed to drive on this road or what type of road are we talking about? Highway or a dirt road? As you can imagine, on narrow, bendy local roads, vehicles often have to drive slower. Similarly, vehicles often need to slow down for turns at intersections, roundabouts, exit ramps and so on. And to make it a bit more complex: a large truck will behave differently across these different types of road than a smaller home delivery van. These properties are then aggregated and used in the ETA calculation algorithm, for every road the vehicle will be driving over to reach its destination.
However, just using a map database isn't enough for calculating accurate ETAs. A map database only provides "static" data. It only updates when, for instance, a new road is added or road properties change, like a maximum allowed speed change on highways. To calculate more accurate ETAs, we need live and time dependent data.
Live data contains data which represents the current state of the road network. It includes currently active traffic jams, congested roads and how much delay they introduce. However, live data changes over time. A traffic jam which is active right now, might be gone in an hour when the vehicle reaches that area. Because we should not use live data in these cases, time dependent data is used instead. This data gives us approximate road properties which change over time. For instance, roads become more congested during rush hour while during off-peak there might be barely any traffic.
Matching reality
During the years we've improved our ETA calculation algorithm a lot, we made it more extensible and have been improving its accuracy to match reality closer and closer.
We won't stop there of course. We are currently investigating how to incorporate "special days" in our ETA calculations, where traffic highly differs from the norm. Days such as Christmas, King's Day and the summer vacation tend to have a serious impact on how much congestion there is on roads. Taking these busy days into account will only further improve the accuracy and while predicting the future will never be 100% accurate, we are excited to see how close we can push towards this 100% in the years to come.
These advanced ETA calculations are used all across our platform: in our TD-Matrix, in the Control Tower to notify the planners on exceptions, in the Store Displays of our clients and in the ETA text-messages that are sent to the end-customers expecting their home delivered groceries.
Want to know more about how we do this magic? Please feel free to contact us.
Gerret Sanders & Bart Toersche
Software Engineers @ Simacan