“Anfangen ist leicht, beharren eine Kunst”
“To begin is easy, to persist is art”
Estimated total eCommerce market on 2012 in Europe is €246 Billion and it’s growing year by year. Its two fold of Indonesia’s state budget 2013 (APBN 2013) and I think this trend will spread out to South East Asia in near future once governments enhance their infrastructures (roads, highway, port, railway, bridges, telecommunication and so on and so forth). The rest is trickle down economy like President Obama said.
As an engineer, I ask a question “what is the heart of eCommerce in modern country like Germany, USA and Japan?“ There’re so many variables but from my perspective: communication, supply chain and capital are the key.
Another question worth to ask is “where will be the competitive advantage of eCommerce company to be addressed in say 5 years later?” Communication? Nein. Capital? Nein. Every big players has big bucks to spend. I must said it’s the supply chain factor.
“Why supply chain?” Typical consumers all around the world are the same, they always checks the price tag first then make a decision buy or not to buy. The lower the price the better, and we all know how the price of a product is determined (Raw Material cost, Manufacturing cost, Distribution cost and other non linear cost).
Next question is, “how to realize this so called supply chain enhancement to eCommerce company profit margin?” This is where engineering works well…hahaha.
In simple model, eCommerce company profit margin fundamentally come from gross margin dollars minus operating margin dollars. Let me put into context per order; gross margin dollar is order size (say $100) – cost of goods ($70) = $30.
While operating margin dollars is: cost to fulfill ($10) + cost to deliver ($10) + cost to support ($5) = $25.
Therefore the profit margin merely $5.
Now imagine, a customer order online a pack of soup with cost $0.89 and the cost to pick and pack in the warehouse is $1. And that’s before the company deliver to the customer! We have to fix this loophole, but how?
The main problem is, there’s no particular material handling tools (ex: forklift, reach truck, conveyor belt, robot arm) to overcome this efficient pick and pack function. Therefore we have to built one, isn’t that what engineers do?
The solution leads to combination of robotic and muti agent systems (MAS) research to provide a material handling tool that will change the paradigm for traditional pick-pack-ship warehouse and significantly improves worker productivity.
But before we depict to the solution, let me briefly explain the drawback of traditional automation approaches as I’ve seen one myself in my former company at Kawasaki, Japan.
1. Costly: price tags for automated material handling systems typically run in the tens of millions of dollars.

2. Long design cycles: most large DC projects takes 12 to 24 months to bring online, in part due to the difficulty of installing and tuning the automation.

3. Inflexible: once installed, conveyors, sorters, carousels, and other systems are difficult and costly to move. They also tend to be inflexible to changes in the inventory mix.
4. Not expandable: few automation systems can be incrementally expanded, which forces companies to buy enough capacity up front to handle several years worth of anticipated growth. This results in excessive capital expenditures for automation systems that run under capacity for several years.
5. Batch processing: attempts to improve the efficiency of the picking task leads to aggregating orders. Once the products are picked in batches, the warehouse employs expensive automation (sorters) to undo the aggregation and break the batches back up into individual orders.
6. Fixed locations: most of the automation systems rely on products being stored in fixed locations, and most warehouse management software assumes the pickable products are stored in a single location. Like in a retail store, a particular width of shelving must be designated to each product, which makes it difficult to adjust the shelf space to accommodate changes in the stocking level. Further, having fixed locations means that the replenishment worker must move the incoming product to that specific location in order to restock the product.
7. Manual reslotting: because of the batch processing and fixed locations, warehouse managers are constantly evaluating inventory locations to keep workloads balanced and the most popular products in the choicest picking locations. Re-slotting requires a lot of manual movement of product from one storage location to another.
So the solution must improve all above drawbacks, and in some cases, eliminate the problem completely.

Above figure is small region of the solution layout. The green cells represent pod storage locations, the orange ovals the robots (with pods/rack not pictured), and the purple and pink regions the queues around the inventory stations. And below picture is the snapshot of the solution workflow.

The benefit from above workflow are:
1. Greater accountability: each order is filled completely by a single individual, improving accuracy and accountability by reducing the number of “touches” on the product.
2. No downstream dependencies: no one worker’s productivity depends on the performance of workers earlier in a sequential process. Instead, each worker’s station is complete and self-contained.
3. No batch processing: in this warehouse, everything is done in real time. An order can literally be filled within minutes of being received.
4. Location-free replenishment: because any station can be used to put product away, the replenishment process is greatly simplified.
5. Adaptive slotting: because the resolution of the allocation of storage is bins rather than the faces of static shelves, the system much more easily adapts to changes in stocking policies. Every product is automatically given just enough pick faces.
6. No single point of failure: unlike a conveyor, if a drive unit fails, it does not stop the whole floor. The rest of the system continues to operate, and most likely there is no noticeable impact on productivity.

7. Rapid deployment: because there is no fixed infrastructure, a 50 station warehouse can be brought online in a matter of weeks rather than months.
8. Spatial flexibility: the systems can accommodate poles, flow into multiple rooms, and handle other oddities of the environment. By incorporating automated lifts, installation can use mezzanines to fill the vertical space.
9. Expandability: if a warehouse needs more capacity, they simply add more pods, drives and stations.
Key to this solution is the multi-agent control system implementation.

Resource allocation is the biggest holy grail as motivation on doing the research to ask such question:
1. Where to store the buckets in the warehouse?
2. Which buckets to bring to which stations?
3. Which buckets to store new inventory?
4. Which stations to assign order to?
5. Which stations to assign incoming inventory?
And I’m working my brain off this issue to fully understand and improve this problem for my personal research all through the year as I’m going to make this for my M.Sc. thesis.
At the moment, there are two approach that I want to investigate such as combinatorial market allocation

and greedy task allocation based on current available research.

I hope I can manage to compile all the necessary thing into a simulated work in java like sample below (run on my Mac OSX 10.6.8).
I personally believe that the future of material handling in eCommerce will be so bright, especially with the latest innovation in semiconductor/superconductor, the sky is the limit to improve productivity in such warehouses.
And as for Indonesia, the logistics cost is just sky rocketing at the moment. Why? The infrastructure is terrible.
I can only hope for the best in logistics costs if only we’ll have at least 10 new bridges in the next 10 years….hahaha. wtf.
Improve or impair, because efficient supply chain contribute to company profit and nation profit.
That’s all for today’s sharing. Wish me luck.
Alles gute.
Horas.
Salam dari anak Medan di Jerman.





















































