Construction and Motion control of a mobile robot using Visual Roadmap By: Harshad Sawhney Guide:...

Post on 26-Dec-2015

217 views 1 download

Tags:

Transcript of Construction and Motion control of a mobile robot using Visual Roadmap By: Harshad Sawhney Guide:...

1

Construction and Motion control of a mobile

robot using Visual Roadmap

By: Harshad SawhneyGuide: Dr. Amitabha Mukerjee

2

Objective

Source

Destination

3

Introduction

• Inspiration From Human Brain.• The roadmap approach, captures connectivity

of robot’s free space.• 3-DOF mobile robot constructed.

4

Construction Of Robot

• Major Components:

Microcontroller Arduino

Wireless module Xbee

Motor driver2 DC motors

Lithium-Ion battery

Image Source: robokits.co.in

Receives Data

UART communication

5

Camera Input

Image Processing

Wireless data transfer

Microcontroller receives

command

µC sends output

Robot Advances

Destination Reached

Flow Chart of Robot Navigation

NO

Stop

YES

6

Image Pre-processing

• 10k images taken.• Background subtraction performed.• Parameters extracted - robot navigation.

Few images from dataset

7

Initial Image Background subtraction

8

Distance Metric Computation

• L2-norm expansion method.• Dist(X, Y) = sqrt(||X||2 + ||Y||2 - 2*||Y’X||)

9

Graph generation

• k-nearest neighbours calculated.• Robot location as nodes.• k=6 taken.• k=10 ; robot jumps larger distance.

10

Nearest nodes to some vertices

11

Shortest path calculation1. Without Obstacles: – Dijkstra’s algorithm used.

Shortest Path Graph

12

Shortest path calculation

1. With obstacles:– Obstacles image extracted.– Compared the image with the dataset.– Nodes and edges removed.– Reduced to no obstacles case.

13

Obstacle Image

Image of environment with obstacles Obstacle image extraction

14

Shortest path calculation

Shortest Path Graph with obstacles in the environment

15

Robot Motion Control

• Feed the nodes.• Camera: Negative closed loop feedback

mechanism.• Reach till destination.• Real Time.

16

Algorithm for controlling robot

• (x, y, Ɵ): Robot current parameters• (x’, y’, Ɵ’): Node parameters• Ɵ : Robot vector angle.• Ɵ1 : Position of robot and node vector angle.

• Step1: Ƒ = | Ɵ - Ɵ1 |• Rotate till | Ƒ | < ɛ• Step 2: Move till | (x-x’)2+ (y-y’)2|< ɛ1

17

Algorithm for controlling robot

• Step 3: Align till | Ɵ - Ɵ’| < ɛ2

• Steps executed in increasing order of priority.• Thus, the camera provides negative feedback

closed loop system.

18

Results

• Robot accurately navigates.• Videos demonstrating robot navigation.

19

Challenges

• Distance metric computation: limits sampling density.

• Real time motion: possibly leading to collisions.

20

Future Work

• Dynamic obstacle avoidance• Update graph first time; use relative changes

in image for future considerations.

21

References

[1] Amitabha Mukerjee, M Seetha Ramaiah, Sadbodh Sharma, Arindam Chakraborty, “The Baby at One Month: Visuo-motor discovery in the infant robot".[2] Joshua B. Tenenbaum, Vin de Silva, John C. Langford, “A Global Geometric Framework for Nonlinear Dimensionality Reduction", 2000.[3] Jean-Claude Latombe, "Robot Motion Planning”, Edition en anglais. Springer, 1990.[4] Choset, Howie, Principles of robot motion: theory, algorithms, and implementations. MIT press, 2005.[5] Seth Hutchinson, Gregory D Hager, and Peter I Corke. A tutorial on visual servo control. Robotics and Automation, IEEE Transactions on, 12(5):651{670, 1996.