Integrated planning of pickers and AGVs in e-commerce warehouse

Integrated robot-human picker scheduling in high volume e-commerce warehouse. Reduced operation cost by 14%

Area of expertise
Operational Excellence
Core strength
Operational optimization
Interaction type
Operational excellence
Optimization software
Data science & analytics
Business process modelling & driver analysis

How an integrated optimization framework can be used for tactical and operational optimization of warehouse operations.

Warehousing systems serve as vital components within supply chains, facilitating the storage and distribution of goods essential to various sectors.

Traditional warehousing relies intensively on manual processes. However, the exponential growth of e-commerce with its very short response times and low operational margins and the scarcity of blue collar workers to perform the operations highlights a great need for effective optimization frameworks to effectively manage and schedule resources within these warehouse systems.

Picking operations within warehouses account for 55% of the total operational cost of a warehouse and, as such, is a perfect candidate to directing optimization efforts.

500+

AGVs

80.000

orders

5-20 %

operating cost

The challenge

A rapidly growing type of semi-automated warehouse consists of a large fleet of robots transporting movable racks of items to picker stations where human pickers pick items composing customer orders. It is a highly integrated picker operations scheduling and robot routing and path planning problem. In addition, priority customer orders require constant rescheduling of plans to meet strict service level agreements.

Approach

We combined techniques from various fields to offer tools for both tactical scenario evaluations as well as operationally running the warehouse.

Tactical tools such as queueing theory and simulation runs were capable of establishing warehouse parameters such as dimensions, SKU-rack allocation,  robot fleet size, headcount size and personnel rostering for dealing with peak loads. High-quality schedules were obtained using metaheuristic algorithms and operationally these could be adjusted using a multi-agent framework to instantly deal with schedule disruptions and rush order arrivals.

In the project, we were able to explicitly take into account uncertainty of order arrivals and picking times with a minimal impact on computation times.

Why does the PEX approach fit so well in this scenario ?

Grounded in research but with a focus to get to actionable results fast, we iterate quickly with the client on these kinds of problems to get the use case correctly defined, data requirements and limitations scoped early and initial results evaluated for their practical implementation potential.

The pre-study for understanding the context  based on Queueing Theory and Simulation resulted in tactical tools for determining optimal fleet size and operator shift optimization. 

 

Results

RESULT 1
5-20% reduction in operational cost

RESULT 2
30% fewer robots required

RESULT 3
In case of disruptions

    • decentralized decision-making algorithms quickly respond to ensure process continuity
    • centralized decision-making algorithm quickly responds to restore efficient schedules and routes

Get in touch!

We would love to hear from you! Feel free to reach out. Our team is always here to assist you. Contact us today and let’s get things moving!

Minderbroedersstraat 36
3300 Tienen
Belgium
People Exponential logo
© 2024 People Exponential.
 All rights reserved.
Minderbroedersstraat 36
3300 Tienen
Belgium
© 2024 People Exponential.
 All rights reserved.

Get in touch!

We would love to hear from you! Feel free to reach out. Our team is always here to assist you. Contact us today and let’s get things moving!

Minderbroedersstraat 36
3300 Tienen
Belgium