Towards Environmental Monitoring with Mobile Robot M. Trincavelli, M. Reggente, S. Coradeschi, A. Loutfi, A. Lilienthal, AASS, Dept. of Technology, Örebro.

Towards Environmental Monitoring

with Mobile Robot

M. Trincavelli, M. Reggente, S. Coradeschi, A. Loutfi, A. Lilienthal,

AASS, Dept. of Technology, Örebro University, Sweden

Hiroshi Ishida

University of Agriculture & Technology, Tokyo Japan

M. Trincavelli

Gas Distribution Mapping Contents

Emerging need for environmental awareness in particular for air quality monitoring.

Investigate the ability to use mobile robots to address this need: Design of a pollution monitoring robot

How performance varies under different environmental conditions

Challenges for existing gas distribution mapping algorithms to cope with “real” and outdoor environments.

M. Trincavelli

Gas Distribution Mapping Contents

Emerging need for environmental awareness in particular for air quality monitoring.

Investigate the ability to use mobile robots to address this need: Design of a pollution monitoring robot

M. Trincavelli

Gas Distribution Mapping Contents

Emerging need for environmental awareness in particular for air quality monitoring.

Investigate the ability to use mobile robots to address this need: Design of a pollution monitoring robot

How performance varies under different environmental conditions.

M. Trincavelli

Gas Distribution Mapping Contents

Emerging need for environmental awareness in particular for air quality monitoring.

Investigate the ability to use mobile robots to address this need: Design of a pollution monitoring robot

How performance varies under different environmental conditions <add pictures of different environments>

M. Trincavelli

Gas Distribution Mapping Contents

Emerging need for environmental awareness in particular for air quality monitoring.

Investigate the ability to use mobile robots to address this need by: Design of a pollution monitoring robot

Investigating how performance varies under different environmental conditions

Investigate whether gas distribution mapping algorithms cope with “real” and outdoor environments.

M. Trincavelli

Gas Distribution Mapping Contents

Emerging need for environmental awareness in particular for air quality monitoring.

Investigate the ability to use mobile robots to address this need: Design of a pollution monitoring robot

How performance varies under different environmental conditions

Challenges for existing gas distribution mapping algorithms to cope with “real” and outdoor environments.

M. Trincavelli

Gas Distribution Modelling

Motivations – why mobile robots for pollution monitoring?

Oil Refinery Surveillance

1

M. Trincavelli

Gas Distribution Modelling

Motivations – why mobile robots for pollution monitoring? Oil Refinery Surveillance

Garbage Dump Site Surveillance

1

M. Trincavelli

Gas Distribution Modelling

Applications Oil Refinery Surveillance

Garbage Dump Site Surveillance

Urban Pollution Monitoring & Tracking air quality monitoring and surveillance of pedestrian areas

communicating pollution levels to technical staff / pedestrians

1

M. Trincavelli

Gas Distribution Modelling Enhance sensor networks by using robots to

provide higher resolution in measurement.

1

M. Trincavelli

Pollution Monitoring Robot

Kernel Based Gas Distribution Mapping

Experimental Setup

Experimental Results

Conclusion and Future Work

M. Trincavelli

Pollution Monitoring Robot

“Rasmus”

Contents

M. Trincavelli

Measure gases with SnO2 gas sensors

Actively ventilated sensor array

M. Trincavelli

Measure wind with a 3D ultrasonic anemometer

2cm/s – 40 m/s range, 1cm/s resolution

M. Trincavelli

Software is Player Based:

Monte Carlo Localization (amcl)

Obstacle Avoidance (vhf+)

Wavefront path planner

Consistent coordinate systems used to ensure

trajectory.

M. Trincavelli

Gas Distribution Mapping

in Natural Environments –

The Challenges

Contents

M. Trincavelli

Gas Distribution Mapping – Challenges

Chaotic Gas Distribution diffusion

advective transport

turbulence

2

video by Hiroshi Ishida

M. Trincavelli

Gas Distribution Mapping – Challenges

Chaotic Gas Distribution

Point Measurement sensitive sensor surface is typically small

(often 1cm2)

2

M. Trincavelli

Gas Distribution Mapping – Challenges

Chaotic Gas Distribution

Point Measurement

Sensor Dynamics

2

M. Trincavelli

Gas Distribution Mapping – Challenges Chaotic Gas Distribution

Point Measurement

Sensor Dynamics

Calibration complicated "sensor response concentration"

relation

dependent on other variables (temperature, humidity, ...)

has to consider sensor dynamics

variation between individual sensors

long-term drift

2

M. Trincavelli

Gas Distribution Mapping – Challenges

Chaotic Gas Distribution

Point Measurement

Sensor Dynamics

Calibration

Real-Time Gas Distribution Mapping changes at different time-scales

rapid fluctuations

slow changes of the overall structure of the average distribution

2

M. Trincavelli

Kernel Based

Gas Distribution Mapping

Contents

M. Trincavelli

Kernel Based Gas Distribution Mapping

General Gas Distribution Mapping Problem given the robot trajectory

Differences to Range Sensing calibration: readings do not correspond directly

to concentration levels

3

),|( :: t1gas

t1gas zxmp

t1x :

M. Trincavelli

Kernel Based Gas Distribution Mapping

General Gas Distribution Mapping Problem given the robot trajectory

Differences to Range Sensing readings don't correspond directly to concentration

levels

chaotic gas distribution: an instantaneous snapshot of the gas distribution contains little information about the distribution at other times

3

t1x :

),|( :: t1gas

t1gas zxmp

M. Trincavelli

Kernel Based Gas Distribution Mapping

General Gas Distribution Mapping Problem given the robot trajectory

Differences to Range Sensing readings don't correspond directly to concentration

levels

instantaneous gas distribution snapshots contain little information about the distribution at other times

point measurement: a single gas sensor measurement provides information about a very small area ( 1cm2)

3

t1x :

),|( :: t1gas

t1gas zxmp

M. Trincavelli

Kernel Based Gas Distribution Mapping

Time-Averaged Gas Distribution Mapping Problem given the robot trajectory

Kernel Based Gas Distribution Mapping interpret gas sensor measurements zt as

random samples from a time-constant distribution assumes time-constant structure of the observed gas distribution

randomness due to concentration fluctuations (measurement noise negligible)

kernel to model information content of single readings

3

),|( :: t1gas

t1avgas zxmp

t1x :

Achim Lilienthal and Tom Duckett. "Building Gas Concentration Gridmaps with a Mobile Robot".

Robotics and Autonomous Systems, Vol. 48, No. 1, pp. 3-16, August 2004.

M. Trincavelli

Experimental Setup

Contents

M. Trincavelli

Experiments

For each environment:

5

M. Trincavelli

Experiments

For each environment:

Introduce an odour source.

Small cup filled with ethanol.

Placed on the ground in the middle of inspected area.

5

M. Trincavelli

Experiments

For each environment:

Introduce an ethanol outdour source

Follow a pre-defined sweep at 5cm/s measuring at stop points every:

10 sec (outdoor)

30 sec (indoor)

5

M. Trincavelli

ExperimentsFor each

environment:

Introduce an ethanol outdour source

Follow a pre-defined sweep at 5cm/s measuring at stop points.

Vary sweeping trajectory from different directions

5

M. Trincavelli

ExperimentsFor each environment:

Create a Gas Distribution Map.

Lighter shaded areas represent higher “concentration”.

Red regions represent relative concentrations levels above 80%.

Blue dots marks the location of measured highest concentration.

5

M. Trincavelli

Experiments

For each environment:

Overlay Wind Measurements.

Arrows coloured according to relative strength from blue to red.

5

M. Trincavelli

For each environment:

Overlay Wind Measurements.

Arrows coloured according to relative strength from blue to red.

Overlay spatial information.

M. Trincavelli

Experimental Results

Contents

M. Trincavelli

INSERT MOVIE CLIP HERE

M. Trincavelli

M. Trincavelli

M. Trincavelli

M. Trincavelli

M. Trincavelli

M. Trincavelli

First Half

Second Half

M. Trincavelli

Initial experiments illustrate:

Difficulties of GDM mapping for real world applications without a ground truth.

The spatial distribution of a gas is unknown

Temporal distribution of the gas

Wind information can provide further clues about the results.

Gas distribution in real environments is a complex problem and this impacts many mobile olfaction applications.

Future work will need to examine the correlation between the instantaneous gas concentration and wind velocity vector in the GDM.