Minimalistic Robot for Mapping and Coverage Supervisors: Dr. Amir Degni Mr. Koby Kohai Students’ names: David Shallom Guy Greenhouse Date: 10/25/2012 Control.

Minimalistic Robot for Mapping and Coverage

Supervisors: Dr. Amir Degni Mr. Koby Kohai

Students’ names: David Shallom Guy Greenhouse

Date: 10/25/2012

Control and robotics laboratory

The Mission

Mapping and reconstructing an unknown map, using a minimal amount of sensors.

So, how many sensors are required for this

task?

Let’s see how Roomba does it!

IRobot Roomba

Roomba’s Structure

Can we do better?

The project probably won’t save livesBut for some, cleaning might be a nuisance.

Progress so Far

Comprehensive market research. Formulation of three major algorithms.Development of a Matlab GUI simulator that

can emulate the robot unique mechanism.Examination of linear and angular errors’

impact on the quality of coverage, and statistics collection.

First steps towards creating the robot.

The Main Algorithm’s InspirationBranch Prediction

The Main Algorithms

Algorithm 1 – simple trend-keeping movement.

Algorithm 2 – partitioning the map into several connected convex hulls using a trend-shifting movement.

Algorithm 3 – beginning with the second algorithm and after stabilizing – continuing with a random movement (Not implemented yet).

The GUI

The First Algorithm

The Second Algorithm

The Third Algorithm (Hypothesis)

Random lengthRandom angle

Angular error [%]

Angular error [%]

Angular error [%]

Angular error [%]Angular error [%]Angular error [%]

Simulation Results – Angular Error

* Each point in the graph represents an absolute deference between the compared parameter averaged over 100 measurements.

Simulation Results – Linear Error

0 1 2 3 4 5 62.5

3

3.5

4

4.5

5

5.5

6

6.5

7

7.5x 10

4 Area Error

0 1 2 3 4 5 60.002

0.004

0.006

0.008

0.01

0.012

0.014

0.016

0.018

0.02

0.022Eccentricity Error

0 1 2 3 4 5 620

25

30

35

40

45

50

55EquivDiameter Error

0 1 2 3 4 5 625

30

35

40

45

50

55MajorAxisLength Error

0 1 2 3 4 5 612

14

16

18

20

22

24

26

28MinorAxisLength Error

0 1 2 3 4 5 60

0.5

1

1.5

2

2.5Orientation Error

Linear error [%] Linear error [%] Linear error [%]

Linear error [%] Linear error [%] Linear error [%]

Simulation Challenges

Numeric precisionBeing able to determine if a certain corner is

convex or concave.

When do we consider the job as done?Map reconstruction based on the robot’s

memory trace.Maps and polygons’ comparison.

Robot Sketch

Bump sensors

Practical Challenges

The main challenge is to develop a reliable mechanism that matches the simulation and theory as much as possible.

Learning how to interface with the Arduino.

Creating a trusty error-immune system.

Being able to compensate the lack of sensors with extra mechanism.

Plans for the Next Semester

Implement the third algorithm in simulation.Dive into statistics collection over a larger database of unknown maps.Continue developing the robot. Configuring the Arduino micro-processor (“the

brain of the robot”).Creating a relevant error model for the specific

mechanism and updating that model in simulation.

Writing an article for IEEE.

Thank You!