Hype and Hoopla About Machine Learning : Logistics Industry Perspective

by | Jun 30, 2020

Artificial Intelligence and its subset Machine Learning have caught attention of many fiction writers and movies since almost 7 decades. There came many phases with lots of promises about the potential of AI and every time it turned from life & world changing technology to almost useless.

Is this time different? The win of DeepMind’s AlphaGo over one of the best players and champion at a game of GO , an ancient Asian game known for its complexity, was the epitome of the superiority of Machine Learning. It heralded the era of unprecedented prognosis spanning from utopia to doomsday scenario.

Few went to the extent that AI will enable humans to enjoy life without work and many have painted gloomy pictures with job losses at scale , threat to human race , inequality etc. Companies like PwC and McKinsey predict that AI will add 10s of trillion dollars to global economy by as early as 2030.

However, the reality lies somewhere in the middle. Bill Gates wrote that we tend to be overoptimistic of any technology in the short run and underestimate its potential in long term.

What may be termed as golden era started in last 2-3 years. It has made umpteen aspects of human life easier. AI and especially its subset Machine Learning ( ML) has provided many tools for businesses to be more efficient and client centric. It has pervaded many spheres of our life.

The investors poured billions in startups with AI ambitions. However, off late the enthusiasm seems to be dampening with cooling of interest from investors and other stakeholders. A prime example is autonomous cars. Google and Uber predicted that autonomous cars would start lining the roads by 2018 and 2019 respectively.

But ground reality is not that rosy with more trials and test underway by both the companies and other companies as well. Since real life is complicated with billions of permutation and combination, the AI techniques are still not upto the scratch to recognize all the patterns . It was easy to reach 90% of accuracy but damn difficult to reach to 99% and will be a herculean task in improving by every fraction of percentage.

 The reason may significantly be attributed to the general technique used in Machine Learning. In almost all the cases, test data is prepared with output labelled by humans . The data is then fed to algorithms and algorithms find patterns. The patterns are thereby utilized to guess the output of unseen data.

So, is it the beginning of another cold winter for AI and Machine Learning ?

If I am asked about the matter concerning the logistics industry , my answer is an emphatic NO. There can be a possibility that full automation of a car may be not possible in the years to come, but the technologies developed are already being deployed for making car driving easier and safe.

Similarly , logistics industry can benefit immensely even if ML squeezes out 10% efficiency from trillions of dollars worth industr . The reason is that costs of developing the ML based tool are a small fraction of the efficiency gains.

In what ways Machine Learning can benefit ? There are few ideas that I will share in this article that can improve efficiencies manifold. To start with Transport Management System can be an immediate beneficiary. Machine Learning combined with IoT can bring automation to entire process namely:

  1. Logistics Plan : Machine Learning Algorithms when connected with your dispatch plan can generate appropriate number and right type of vehicles for each origin and destination on a given day.
  2. Indentation : It can automatically communicate the indent to all the service providers ( transporters)
  3. Vehicle Placement: No need to call transporters as transporter may be reminded for non-placement automatically
  4. Checks : Once the vehicle arrives at the origin it is automatically allowed in the loading . ML can check the bona fides automatically and talk to IoT machines to allow the vehicles.
  5. Within Plant Tracking : It can provide a real time view on vehicle location and generate alerts if there is a delay . It also provided bottleneck analysis on real time basis. It will automatically calculate efficiency on each stage to take preventive and corrective actions on waiting time, loading time, etc.
  6. Document generation : ML will allow you to automate generation of documents like E-Way bill, invoices, POD
  7. Tracking : IoT an you real time tracking with Estimated Time for Arrival ( ETA) and alerts in case of delays with the constant update of ETA
  8. Delivery : Proof of Delivery Document may be generated automatically when the consignee / receiver gets the consignment with proper checks & dispute notification
  9. Accounting entries : It can allow you to update all the accounting entries automatically in your ERP or Accounting System

The above may sound  a part of some fiction, however, many of these are already developed or underway.

The good news is that users don’t need to invest heavily in developing these systems and could hitch the ride immediately with instant advantages.

Few points of caution while evaluating any such system are:

  • The system should be comprehensive enough to incorporate the entire process or at least easily integrable with existing systems
  • The system should be able to integrate IoT in seamless ways. This is one of the major issues that most of the systems are developed in isolation to solve single problem and thus are not able to integrate well when there is need.

Conclusion

The emergence of AI and its subset Machine Learning (ML) has a long term potential for all businesses especially for the logistics and transport industry. The logistics industry is marred by :

  • Lack of transparency
  • Human involvement
  • Low quality and incomplete data
  • Almost negligent analytics to improve business

It is worthwhile for the users of Transport services to automate the entire workflow from Dispatch planning to Delivery. It will avoid costs of mistakes and delays with accurate and automatic data collection, document flow  and information sharing. Third party tools are available to convert at least a significant part of the Transport Management System.

Companies are well advised to commence, at the very least, considering Machine Learning as a tool which if implemented can do wonders and if not can even pose existential threat.

 

Contact us :

Rohit Chaturvedi
CEO