Optimisation of laser data acquisition: Robots and lasers
1. Optimisation of laser data acquisition
Robots and lasers
Marek Ososinski
Aberystwyth University
Marek Ososinski (Aberystwyth University) Robots and lasers 1 / 18
2. Layout
1 Introduction
2 Project Desription
3 System overview
4 Environment Mapping
5 Reasoning, Acquisition and Validation
Marek Ososinski (Aberystwyth University) Robots and lasers 2 / 18
3. About me
Name: Marek Ososinski (mro7@aber.ac.uk)
Place: Aberystwyth University
Background: BSc in Artificial Intelligence
Funding: KESS
Partners: Royal Commission on Ancient and Historical Monuments of
Wales
Marek Ososinski (Aberystwyth University) Robots and lasers 3 / 18
4. A bit of history
Marek Ososinski (Aberystwyth University) Robots and lasers 4 / 18
5. Project Description
Project:
The aim of the project is to provide a methodology for laser data
acquisition, that will esnure best quality of data within the given
time-constraints.
PhD:
Creation of a robotic system capable of mapping the area and
detecting best scanning positions, that will ensure completeness and
minimise registration error.
Marek Ososinski (Aberystwyth University) Robots and lasers 5 / 18
6. General idea
to here
getting from here
Marek Ososinski (Aberystwyth University) Robots and lasers 6 / 18
7. Optimisation of laser data acquisition
Optimisation factors:
Time
Registration error
Scan completeness
Processing effort
Marek Ososinski (Aberystwyth University) Robots and lasers 7 / 18
8. System Overview
Four stages:
reconnaissance
reasoning
acquisition
validation
Aims to be autonomous, might require human supervision.
Marek Ososinski (Aberystwyth University) Robots and lasers 8 / 18
10. Reconnaissance stage - Robotic platform
stable platform
limited indoor access
needs pitch and roll
corrections on uneven
terrain
cannot traverse stairs
internal power supply
robotic platform
Marek Ososinski (Aberystwyth University) Robots and lasers 10 / 18
11. Reconnaissance stage - Organic bipedal platform
unstable platform
requires constant pitch
and roll corrections
go-anywhere ability
requires external power
supply for equipement
allowed inside historical
monuments
runs off chips and coffe
organic bipedal
platform
Marek Ososinski (Aberystwyth University) Robots and lasers 11 / 18
12. Environment Mapping - Occlusions
Marek Ososinski (Aberystwyth University) Robots and lasers 12 / 18
13. Environment Mapping - Occlusions
Occlusion detection types:
from single viewpoint using movement
multiple viewpoints using optical flow
multiple viewpoints using object correlation
Assumptions:
static environment
known position of the viewpoint
Marek Ososinski (Aberystwyth University) Robots and lasers 13 / 18
14. Reasononing stage
Detection of laser scanner positions
User supervision:
provide the desired resolution based on map visualisation
provide optimisation parameters
confirm that scanner positions are accessible
Automation:
based on optimisation parameters
using the occlusion map
Marek Ososinski (Aberystwyth University) Robots and lasers 14 / 18
15. Acquisition stage - Robotic platform
Acquisition:
Use of HDS6200 laser scanner
Robotic platform will traverse to
the accessible scanning locations
Fast
Stable platform
No supervision
Limited accessibility
Marek Ososinski (Aberystwyth University) Robots and lasers 15 / 18
16. Acquisition stage - Human
Acquisition:
Use of HDS6200 laser scanner
Human operator will place a
tripod in less accessible locations
Requires manual tripod leveling
Can get to the hard-to-reach
places
Marek Ososinski (Aberystwyth University) Robots and lasers 16 / 18
17. Validation stage
Validation:
Detection of potenetial registration targets
Estimation of registration error
Estimation on registration completeness
Marek Ososinski (Aberystwyth University) Robots and lasers 17 / 18
18. Finally the last slide !
Any questions?
Marek Ososinski (Aberystwyth University) Robots and lasers 18 / 18
