Technion - Israel Institute of Technology
Electrical Engineering Faculty
Control & Robotics Lab
Optimal Area Covering by Mobile Robot
presented by: Nelly Chkolunov
Fentahun Assefa-Dawit
supervised by: Alexey Talyansky
Introduction
Mobile Robot Indoor Applications
• Servicing equipment
• Servicing humans
• Other Functions
Covering Problems:
• off-line
• on-line
Project Description
• Study of the main covering algorithms:
- deterministic
- probabilistic
- semi-probabilistic
• Compare these algorithms
• Choice one of them and apply on the robot
Tools
The Pioneer 1 Mobile Robot
Key features: Reliable
Portable
Plug’N’Play
Versatile
Supported
Software Support
Saphira (version 6.1)
Pioneer Application Interface (PAI)
Visual C/C++ env.
Communication Devices
Radio Modem Pair
Related Papers
• I.A. Wagner, M. Lindenbaum and A.M. Bruckstein,
“MAC vs. PC - Determinism and Randomness as
Complementary Approaches to Robotic Exploration of
Continuos Unknown Domains”
• B. Kuipers and Y. Byun,
“A Robot Exploration and Mapping Strategy Based on a
Semantic Hierarchy of Spatial Representations”
• C. Hofner and G. Schmidt,
“Path planning and guidance techniques for an
autonomous mobile cleaning robot”
MAC (Mark And Cover) the Deterministic Algorithm
• finds an uncovered point around the current location
• marks the covered points
• time needed to cover
region with area A:
T
MAC
 O A / a 
MAC (Mark And Cover) (cont.)
Advantages:
 guaranteed coverage of a connected region
 awareness of completion
 very efficient on not too large regions
Drawbacks:
 vulnerable to physical problems (e.g. sensory
errors, dependence on a memory)
 not efficient for a large regions
PC ( Probabilistic Covering)
• randomized
algorithm
• make a short step and then random turn
• average cover time:


E T PC  O Ad a 
PC (cont.)
Advantage:
 almost sensorless
(just sensors to identify a collision)
Drawbacks:
 the average performance is lower than MAC
 no awareness of completion
MAC-PC: A Hybrid Algorithm
• a combination of two first:
works as PC globally and
as MAC in local steps
• expected coverage time:

ET
MAC  PC

 A
 A

 r ' r log  r '2






MAC-PC (cont.)
Advantage:
 reasonable tradeoff between the performance
of the first method and the robustness
of the second one
Drawback:
 still no guarantees a complete coverage
Our Solution
MAC-PC - Semi-Probabilistic Covering
1. Cover local area by applying MAC
process from current location
2. Choose a random neighbor
from a local area boundary
3. Go there
4. Repeat all mentioned above
number of times expected for
complete coverage of the total area
Our Solution (cont.)
Local Area covering (MAC rule):
• Preprocessing stage:
Sonar check
Local area bounds check
• Processing stage:
Backtracking in form of milling
• Robot trace stored in data base
during the local area covering
Our Solution (cont.)
• Expected Number of Random Steps :
• Simulations & experiments:
assumptions and problems
Summary
• We studied the area covering problem
and its different solutions
• Applied one of them (MAC-PC)
on the Pioneer 1 mobile robot
• Ran the various tests