1 of 27

Puzzlebots

Team a: mind the gap

2 of 27

agenda

2

1.	Team introduction
2.	Project background
3.	Needs analysis
4.	Objective tree
5.	system requirements
6.	Functional Architecture
7.	Looking ahead

3 of 27

Team

Introduction

3

4 of 27

4

Sankalp

Mech Wizard

Hari

Integrator

Sandhya

Controller

Sudhansh

Visionary

Dhanvi

Man with the Plan

5 of 27

Project background

6 of 27

Puzzlebots today

6

1.

2.

3.

4.

Our job?

To scale this up!

7 of 27

What does scaling up change?

7

We require a new coupling mechanism to bear additional weight
We only want to use on-board sensors
We want the system to determine the appropriate configuration of robots
We need to redesign the coupling algorithm to fit our mechanism

8 of 27

Needs

analysis

9 of 27

WHY?

10 of 27

Use case development: the journey

Typical process

Our process

10

11 of 27

How would you do it?

12 of 27

Use case development: solution scoping

12

Solution	Variable Length	Jumping	Drones	Coupled Swarm
Favourable	Simple Controls	Fast No constraints on the gap	Fastest No constraint on gap Self sufficient	No constraints on gap LOW work volume Variable configurations
Unfavourable	Weight Distribution Maintaining Balance Life Cycle Constraints on gap	Vertical Clearance Difficult to control Mechanical wear and tear Payload constraints	Vertical clearance Susceptible to weather Inefficient for ground tasks	Slow Difficult to coordinate

13 of 27

“Autonomous collaborative ground traversal of discontinuous terrain”

Ex: Data collection from inaccessible locations like caves/tunnels

13

14 of 27

Objective

tree

15 of 27

15

16 of 27

functional requirements

17 of 27

Functional requirements

17

F1.	The system shall localize agents in a given map
F2.	The system shall route agents and avoid collisions
F3.	The system shall sense gaps and determine their feasibility
F4.	The system shall determine and achieve coupled configurations
F5.	The system shall cross gaps
F6.	The system shall reach given regions of interest

The system should be able to localize the agents first. Next based on the given points of interest, the system shall route agents towards the points of interest and also avoid collisions. Once the individual agents starts moving around, the system should be able to sense any gaps and determine if crossing is feasible. Once a feasible gap is found, the system should determine the optimal configuration required for crossing and the agents shall achieve the coupled configurations. The coupled system shall then cross the gaps. Finally, on a higher level, the system shall reach the given points of interest initially given.

Okay, I just listed down the functions that we want to demonstrate by the end of next year. But we do not actually have any numbers to evaluate our success here. To address that, we listed down our performance requirements.

18 of 27

PERFORMANCE requirements

18

P1.	The system will deem gaps upto 1.5 agent lengths feasible
P2.	The system will have 0 unplanned collisions between agents
P3.	The system will achieve formations with at least 3 robots
P4.	The system will cross feasible gaps 75% of the time
P5.	The system will reach all POIs 75% of the time
P6.	The coupling mechanism will bear the weight of one agent

The performance requirements mentioned here have been elicited either from discussions with our sponsor or from the original Puzzlebots system which we are trying to scaleup. The system will cross gaps of length upto 1.5 times the individual agent length. We want the system to avoid unplanned collisions 100 percent of the time. For those who are wondering why would someone in their right mind, plan a collision among the agents, the planned collisions are basically the coupling behaviour. Next, the system will achieve formations involving atleast three robots. We can couple just two robots and say we achieved coupling but we would like to go for coupling between more than two robots. Crossing the gaps being one of the main novelties of the system, we would want it to be robust enough to cross feasible gaps 75 percent of the time. The system will reach all the given points of interest 75 percent of the time. The coupling mechanism that we design should bear the weight of atleast one agent. Now that we have seen the performance metrics we would like to achieve, there are some more non-functional requirements that we wanted to address.

19 of 27

Non-functional requirements

19

N1.	The weight of an agent shall be minimal
N2.	The coupling mechanism shall consume a low amount of energy
N3.	The system shall be expandable
N4.	The system shall be easily maintainable
N5.	The team shall maximize learning and fun throughout the project

20 of 27

functional Architecture

21 of 27

functional Architecture: System

21

22 of 27

functional Architecture: server

22

23 of 27

functional Architecture: AGENT

23

24 of 27

Looking

Ahead

25 of 27

What does success look like?

Before

After

25

26 of 27

Roadmap

26

1

3

5

6

4

2

Refine requirements with sponsor

System-level trade studies

Testing capabilities of Khepera platform

Cyber-physical Architecture

Subsystem-level trade studies

Finalize Spring Validation Demo scope

27 of 27

Questions ?