Break the Mold with Real-World Logistics AI and IoT
We have been talking a lot, lately, about the Internet of Things (IoT) and Artificial Intelligence (AI). So much so that it’s now difficult to differentiate the real from the not-so-real or purely ‘marketing’ IoT and AI. Data mining isn’t AI. Marketers have been doing it for a good three decades, and others likewise. It’s using intelligent correlations and cohorts to find patterns and latent needs. That’s not much that is artificial about the issue nor situation.
There should be a new marketing codebook with these lines: “Thou shalt not cite IoT and AI in vain.” I don’t know how, but the salesperson calls my latest watch “AI enabled,” whether they have AI or not. The clock is not even smart; at best, it’s just digital. When you wipe off the not-so-real jargon and look at the actual applications of AI and IoT, they are aplenty. But how do we find what is actually true — in a world so taken with these terms? It’s simple.
Just know the story behind the pitch. Does the product or solution improve over time? In a customer-facing scenario, does it customize itself to your language (maybe like the Amazon Echo).
In a more enterprise setting, does it offer better/faster delivery routes for your logistics movement each time you use it? Does it incrementally better itself with a singular goal of improving the results, learning and adjusting? If yes (to any), then it’s AI.
A system which learns on itself and tells right from wrong;
A recent use-case comes to mind. The company I am associated with, LogiNext, used Kalman filters (algorithm). NASA made the Kalman filter famous when they used the algorithm in their effort to better direct satellites in near and outer space. According to a paper, right back from 1985,
“The Kalman filter in its various forms has become a fundamental tool for analyzing solving a broad class of estimation problems.”
The company in question used an updated iteration of the Kalman filter to fix vital tracking information of hundreds of trucks moving across the country. Hence, each tracking point was, then, accurate up to 3×3 yards. What’s the impact?
- Precise knowledge of where each truck is located.
- Where the truck will be in the future.
- And when this vehicle will reach the destination; down to the minute.
The updated algorithm, with the layer of Kalman filter, learns from the tracking errors. It is essential as the tracking is hardware and network coverage dependent. It identifies patterns in the tracking data to understand what is ‘credible’ monitoring and what’s an error. The system would itself know which tracking data to use and which to ignore, growing the accuracy with continued functioning.
In turn, this would ensure that the information going into the system for processing and route planning is accurate. More importantly, avoiding another case of ‘garbage in, garbage out.’ It would be more consistent with incrementally better plans each time it’s used.
Here’s the IoT you can use, with complete logistics streamlining.
Logistics is primarily a game of Service Level Agreements, SLAs. A company/carrier needs to adhere to these basic unit agreements, SLAs, or minimum viable service levels. It may be when a shipment leaves, the quality of the truck or environment for the cargo, the time when it needs to reach, etc. These SLAs are the code of conduct for carriers, drivers, and companies. They are specific to each shipment. SLA breaches are a serious affair and may result in delays and eventual penalties.
So, with SLAs at the center stage, when you must track a package from perhaps LA to NY, you would expect a continuous flow of information regarding the location and state of your package, along with tracking the adherence to the all-important SLA, the ‘promised delivery time.’ How is your estimated time of arrival