Project Group 7: MANS-i

Objective

Our team is currently interested in implementing a virtual swarm robotics style maze solving system where the addition of more robots optimizes completion time. Robots will be restricted to ninety degree turns and work together by wirelessly communicating with a central database. This central database will ensure tasks are not duplicated and robots do not collide with one another. We will use the resulting data to generate a large set of widespread simple maze statistics that collectively converge to an ideal relation between maze complexity
and robot swarm count.

Mentor

Professor Richard Lathrop, ICS

Team Members

• Nathan Le, CpE
  • Position- Software Manager
• Steven Chow, CSE
  • Position- Hardware Manager
• Andrew Sperry, CSE
  • Position- Vision Manager
• Mansi Tyagi, CSE
  • Position- Team Manager

Introduction

Swarm robotics is defined as a field of multi-robotics where a large number of robots work together cohesively in a distributed and decentralized way to solve complex scenarios or tasks. Each robot follows a set of local rules that are simpler than the globally complex objective. Currently, there is a wide variety of design implementations that have been developed to solve mazes. These include left or right turn only algorithms, depth-first search, breadth-first search, last-in-first-out, random mouse, wall follower, shortest path, and many others. Our team hopes to expand on the existing research by providing an ideal relation that determines the optimal robot swarm count to use given a particular style or complexity of maze.

Schedule

Fall Quarter

• 1st Half
  • Identify the needed hardware and items for the project.
  • Complete the Maze UI specification Sheet
• 2nd Half
  • Begin coding the Robot AI
  • Begin Hardware Setup to support the software implementation.

Winter Quarter

• 1st Half
  • Have a single robot setup and ready to begin testing. 
  • Begin simulations to collect data and identify bottlenecks and areas of improvement
• 2nd Half
  • Expand to have multiple robots up and running at the same time.
  • Have a UI with real-time stats displayed for spectators to see what is happening.

Spring Quarter

• 1st Half
  • Begin construction of the physical version of the robot.
  • Refine the maze-solving algorithm to improve efficiency and reduce bottleneck
• 2nd Half
  • Construct a physical maze that can be changed on the fly to highlight the robot’s adaptability.
  • Construct multiple version of the physical robot.

Diagrams

System Design

• We will run the simulation on a Raspberry Pi that will output the UI via HDMI

UI Design

• The UI will display the statistics of each maze. The main panel will display the maze and the robots as they move through it.

Results: Projected Statistics