Location intelligence with geospatial data
Feb 11, 2020 Comments (0)
Companies which ship products to customers know the importance of velocity, visibility & efficiency.
Location Track and Trace using GPS capabilities is now a fairly mature technology to the point that it is commoditized. While real-time location of a shipment or an asset is an important data point for Logistics Service Providers (LSP’s), Shippers, Consignees et al, the personalized intelligence that can be derived from such data along with self-orchestration capabilities is the real value proposition.
Advancements in Edge computing, Analytics and Machine Learning technologies now make it possible to rise a step up in the value chain and provide insights which can predict, prevent and prescribe events and actions.
We focus on how three areas; Shipment Dwell time, Geofencing & Vehicle Expected Time of Arrival (ETA) can uncover latent business value to the logistics ecosystem.
Dwell time Intelligence
A vehicle traversing a certain number of routes over a period of time gathers data on transit times and waiting times along the route. The waiting time could be during pickup, delivery, enroute at a fuel or rest stop, at a transhipment terminal, port, or even a traffic jam. Each waiting location has its own characteristics and each unscheduled or unplanned wait has a cost associated to it.
Logistics is all about efficiency and speed and many-a-times it is not possible to pass on such dwell time costs to the customer. An intelligent analysis of dwell times at various locations along with other datapoints can provide insights on the best approach to handle a current or planned shipment along the same route.
Lets take an example of a Full Truckload Delivery at a customer warehouse. The time a truck spends in the customers premises for waiting and unloading is easily known and based on past data analysis an ‘average’ waiting time can be calculated. However if this base information is coupled with other datapoints like truck arrival time, day of the week, trailer size, commodity type, number of shipments delivered, packaging type, deviation in minutes from original appointment window etc. then a combined analysis of all these factors can present a much finer view on dwell times. Two trucks of the same LSP arriving at the same customer warehouse can likely face varied dwell times depending on the combination of above parameters.
A prediction of truck dwell time based on a machine learning model with above features can not only help in knowing the individual truck turnaround time from the customers premises but also serve as a very useful input to plan the next trip. If a modification in the Delivery appointment window for the next delivery is warranted due to excess dwell time predicted in the current delivery then this self orchestrated action can be initiated much before. Thus a cascading effect of inefficiencies can be mitigated through refined intelligence.
Take another example of traffic congestion. A careful analysis of past truck movements along different routes along with other data points like trailer size, time of the day, day of the week, weather, vehicle speed, idling time etc. can yield insights into location hotspots where a truck spends maximum time waiting for traffic to move. The truck could either be stationary – stuck in a traffic jam – or moving at a very slow speed in a given area. Historically if a similar pattern of slow movements is observed for certain areas then this information can be used for route replanning of future shipments. This can be used to derive more realistic transit times. This can be used to alter departure times.
According to a report by the U.S. Department of Transportation, a 15-minute detention period can increase the expected crash rate of a vehicle by 6.2 percent.1 Drivers are always under pressure to ‘make up’ for the extended waiting time at a premises by overspeeding and ‘be-on-time’ for the next scheduled delivery. Rash driving can of course result in accidents and it is this safety aspect which also warrants refined predictions of dwell times at various locations for a drivers journey. A realistic schedule which takes inputs from past learnings of dwell times can help reduce driver fatigue and avoid road mishaps since the days trip would have factored in the precise waiting time predictions and arrival times at various stops.
One of the core and distinct functions of the Geo-Spatial Intelligence is the ability to leverage geo-fences for avoiding an unsafe situation of a shipment. Traditionally geo-fence functionality is used to just report the time that a cargo or an asset has spent in the geo-fence. Some of the functionalities also include providing warnings before an asset or cargo is in the vicinity of a negative geo-fence. But its time that the commodity nature of this function gets augmented.
Geo-fence functionality powered by Edge Computing at device and cloud can bring the ability to safe guard the condition of cargo in the new age requirement. Triggering edge algorithms based on the geo-fence characteristics (nature of the location, time of the entry, duration spent etc), cargo characteristics (type of the product, current health state of cargo, cargo dimensions, weight etc) and few other external parameters bring a inconceivable foresight on the next and likely state of cargo such that timely action can be taken.
This new age function is developed with a deep understanding of the data generated and modelling it with machine learning methods to be activated at apt trigger points.
Further the nature of applying geo-fences is to be radically re-designed; the way of defining circular and polygon geo-fences served its purpose but doesn’t solve the current day’s logistics industry challenges. A costly cargo movement needs geo-fencing for the entire path that coins a nature of geo-fence called ‘Ribbon’. Similarly the type of functionality in a geo-fence needs an enhanced treatment such a ‘Spot’.
This new age geo-fence function solves a myriad of business challenges in the logistics industry such as temperature excursions, theft possibilities, misroutes, improper cargo handling, unwanted wait times, missing cargo, damages, shortages and many others. A quick list of capabilities that the geo-fencing enables are listed below:
· Avert misroute by binding the shipment to geo-fenced area
· Ensure honoring appointment windows and avoid consignee penalties
· Enable the way to create the data for accessorial realization (using spot geo-fence)
· Monitor shipments missing their schedules (pickup, delivery, sorting, line hauls etc.,)
· Assist in the way of averting a theft or pilferage of shipments
· Avoid traffic bottlenecks (avoid negative geo-fences) on the route and save the transit time
· Enable co-relating with the status of other shipments
· Predict arrival pattern of shipments at terminals /hubs/cross docks for activity planning
· Auto-Geofence creation: Personalized, ad hoc, shipment based geofences creation.
The ability to accurately predict vehicle arrival time at a location is not just an option but a key differentiator for LSP’s to implement optimization strategies, boost customer satisfaction & to have a profitable bottomline.
The sooner a customer knows how & when a shipment will arrive and whether it will affected by any hiccups,the more leeway there is to look for alternatives so as to ensure smooth operations .
Vehicle ETA calculation seems fairly commonplace. But once you start going deeper into the shipment details, you realize that not all ETA’s are equal. The vehicle transit time depends upon certain unique factors like vehicle size, vehicle type, no. of trailers attached, the type of products being carried, the planned route etc. A much refined ETA prediction is thus possible and it is this which brings efficiencies to the ecosystem.
Real-time alerts coming from such a system could alert the stakeholders about delayed shipments and provide proactive re-planning options such as bringing in alternative shipments ,rerouting shipments to another dock, ensuring personnel’s are in the right location at the right time ,alerting other business units about the possible shift in operational priorities et al.
As per a Gartner report, the number of organizations using location intelligence will grow four-fold by 2021. Location Intelligence Market Study conducted by Dresner Advisory Services highlights that 66 % of the enterprises interviewed, believe location intelligence to be decisive to improve their revenue & growth strategies.
Geospatial Location Intelligence powered by edge, AI and cloud can now make cargo speak for itself. It can provide a voice to every single piece of shipment or asset and self orchestrate actions to any predicted exceptions. The endgame is to achieve optimized operations, increased bottomline and last but not the least great customer satisfaction.
The paper has been authored by Pradeep Chaudhary, Srinivas Patchala and Avani Gangale who are part of the Transportation Strategic Initiatives team at Tata Consultancy Services Ltd.
Pradeep Chaudhary is a Domain Consultant with Tata Consultancy Services - Travel, Transportation and Hospitality business unit. He has over 20 years of experience in the logistics and supply chain area. He works for the Transportation Strategic Initiatives group and is responsible for creating innovative digital offerings for TCS’ clients in the logistics space. Chaudhary has worked on multiple projects…