1 of 1

Results

  • A map plot of the driver, restaurants and customers location on map
  • Assignment of driver to a order
  • Reduced Makespan
  • Optimized routes and total travel times and distances
  • Orders are delivered in sequence as per the optimized route generated, not necessarily at once

​

Plot of points:

​

​

​

Plot of routes:

Case1:

4 drivers, 5 orders and 4 restaurants→ 2 routes

Case2:

3 drivers, 5 orders and 4 restaurants → 3 routes

​

​

Motivation

  • Integrate state-of-the art mechanism to enhance delivery system
  • Fast assignment of drivers to order efficiently�1) with lowest distance�2) time efficient
  • Using Taipy built an optimized efficient delivery system

​

Experiment

​

​

​

​

​

​

​

​

​

​

  • Objective was to minimize the overall distance travelled by each vehicle
  • Built a Two Stage optimizer
  • Use of VRP tools ,Google API and openrouteservice api to solve the problem
  • Created Routes based on number of vehicles

​

Following cases considered:

  • Single order (k) – driver(m) mapping
  • Multiple order (k) – driver(m) mapped based on:
  • Vehicle capacity constraints

The capacity of vehicles is assumed to be 10 in our case. This case is usually applicable when rider has to pick up multiple deliveries along a route

  1. Order Pick up and delivery locations

Restaurants are the pick-up points while customers are delivery locations. As per the algorithm they are randomly mapped. Multiple orders can be assigned to a single rider.

  • Number of vehicles

Different number of vehicles create equal number of routes or less. Number of routes are based on optimized routes.

Assuming the speed is constant, time is optimized automatically

Based on all constraints mentioned above, following are the results:

​

​

​

​

​

​

​

​

​

​

Conclusion

  • Quick implementation, effective and efficient way to leverage technology
  • Use of Google APIs and Taipy integration to solved VRP within short time span
  • Enhance model performance with use of OR tools and Taipy
  • Based on conditions routes can be optimized

​

Future Scope

  • Feedback loop that optimize the next assignment of driver to an order
  • Upscaling the model performance with large number of runs
  • Inclusion of additional risk factors or critical variables to consider

http://www.stevens.edu/bia

Developing Machine learning and Supply chain Application using Taipy

(Python Open Source Advanced Tools)

Rahul Varma Sikinam,

Instructor: Alkiviadis Vazacopoulos

​