top of page

INFORMATION

Client

Professor Sean Anderson

Location

Boston University, MA

Date

1/13/25 – 5/20/25

Teammates

Nathan Lau

Azamat Abdikarimov

Mitchell Hornack

Programs Used

MATLAB

Python 

Visual Studio Code

Screenshot 2025-09-03 at 1.51.55 pm.png

The demand for greater efficiency in the processing and storage industries has accelerated the development of robotic automation. Large corporations increasingly rely on fleets of autonomous mobile robots to transport inventory within their warehouses, gaining significant logistical advantages. Companies such as Amazon have acquired specialized robotics firms, including Kiva Systems and Cloostermans, to build extensive in-house robotic operations under the Amazon Robotics brand. However, despite rapid advances at the top end of the market, access to automation remains highly uneven across the industry. Smaller storage facilities and independent warehouses often struggle to adopt robotic systems due to high upfront investment costs and the difficulty of reconfiguring limited floor space for traditional automation architectures. This growing technological gap risks creating monopolistic market conditions, restricting innovation and widening the efficiency divide between large and small enterprises. To address these challenges, more flexible and scalable robotic solutions are needed. One promising approach involves the use of smaller, lower-cost robots operating collaboratively in a swarm to transport larger or heavier inventory items. By distributing load-sharing across multiple robots, such systems offer better adaptability to constrained environments and reduce the cost barriers associated with traditional large-scale automation. Advancing coordinated multi-robot control strategies is therefore a critical step toward democratizing access to warehouse automation and ensuring that efficiency gains are broadly shared across the storage and logistics industries

Single Leader MPC Method

Dual Leader MPC Method

One robot acts as the sole leader following their path unimpeded, while all other robots follow their given path but adjust speed and trajectory relative to the leader to maintain formation.

Screenshot 2025-09-03 at 1.52.20 pm.png

Both the left and right front robots serve as leaders of their respective side of the square, following an offset path from the centerline while also adjusting speed and trajectory to keep in formation with each other. In this strategy, the rear robots are following the same paths but adjust their controls based only on their respective side leader to keep formation.

Screenshot 2025-09-03 at 1.52.31 pm.png
Screenshot 2025-09-04 at 11.40.00 am.png

3 follower robots penalized for x, y, and yaw deviations from the leader

Screenshot 2025-09-04 at 11.40.09 am.png

Rear robots penalized for x and y deviations from respective leader

Screenshot 2025-09-04 at 11.40.17 am.png

Front robots penalized for y deviations from each other

Single Leader MPC Simulation Results

Screenshot 2025-09-04 at 11.45.28 am.png
  • Excellent trajectory tracking on straight path

  • Slight lateral deviations during turn; formation stretches

  • Robust formation recovery observed after turn

  • Curved path causes larger deviations due to frequent heading change

Dual Leader MPC Simulation Results

Screenshot 2025-09-04 at 11.45.37 am.png
  • Very accurate trajectory tracking on straight and curved paths

  • Better formation stability across straighter paths

  • Worse formation integrity on sharp turn

  • Longer time to reach goal

bottom of page