Volvo Technology Humans System Integration Volvo Technology Effects of visual and cognitive load on...

Volvo TechnologyHumans System Integration

Volvo Technology

Effects of visual and cognitive load on the Lane Change Test – preliminaryresults

Johan Engström &Gustav MarkkulaVolvo Technology Corporation

Driver Metrics WorkshopOttawa October 2-3, 2006


Volvo Technology

Background

• Existing results from (HASTE and other studies) show that visual and cognitive tasks have rather different effects on driving• Effects of visual tasks

• Reduced lateral control (increased standard deviation of lane position)

• Reduced event detection performance (e.g. PDT)

• Speed/headway compensation

• Effects of cognitive tasks• Improved lateral control (reduced variance of lane position)

• Gaze concentration towards road centre

• Reduced event detection performance

• No speed/headway compensation

• Visual and cognitive tasks both result in increased LCT normative path deviation – but could this common effect have different origins?

• Discussed at LCT TF meeting in Shonen Village in connection to reference tasks…


Volvo Technology

Visual vs. cognitive tasks - definitions

• Purely visual=requires diversion of gaze from the road centre but no cognitive activity (e.g. just looking back and forth…)

• Purely cognitive=does not require gaze diversion (e.g. phone conversation, speech interaction etc.)

• In practice, most visual tasks involve a cognitive component – however, the degree of cognitive load varies (e.g. for the Arrows and Circles tasks the cognitive load could be assumed to be minimal)

• However, many real tasks could be regarded as purely cognitively loading


Volvo Technology

Hypotheses tested

• Visual tasks mainly lead to reduced lateral control during the lane changes but also (to a lesser degree) to reduced detection performance

• Cognitive tasks mainly lead to late or erroneous detection

• Thus, both tasks lead to increased deviation from the normative path, but (partly) for different reasons


Volvo Technology

Existing work: Mattes et al. (in AIDE D2.2.5, Östlund et al., 2005)


Volvo Technology

Results, Mattes et al. (in AIDE)


Volvo Technology

Method

• Same data as used by Mattes et al. in AIDE

• 30 subjects (subset of data in the study)

• No task priority in instructions

• Sign: content pop-up

• Secondary tasks:• Cognitive easy: Count up by 2

• Cognitve hard: Count down by 7

• Visual easy: ”Ring detection task” – large difference between rings

• Visual hard: ”Ring detection task” – small difference between rings


Volvo Technology

Results: First qualitative analysis, example 1

= detection/interpretation

errors

= loss of control

= Baseline

= Cognitive (easy +hard)

= Visual (easy+ hard)


Volvo Technology


errors

= loss of control

= Baseline





Volvo Technology

Results: First qualitative analysis – manual classification of errors

Detection/Interpretation errors

0

5

10

15

20

25

30

Baseline Cognitive Visual

Control errors

0

2

4

6

8

10

12

14

16

18

Baseline Cognitive Visual


Volvo Technology

First conclusions from qualitative analysis

• Qualitative analysis seems to support the hypothesis:• Visual task mainly leads to errors related to reduced control, but also

some detection errors

• Cognitive task almost only lead to detection/interpretation errors

• But can these effects be quantified?


Volvo Technology

Preliminary results of quantitative analysis


Volvo Technology

Quantifying loss of control (I): Mean difference of lateral position (~mean lateral velocity)

0.006

0.0065

0.007

0.0075

0.008

0.0085

0.009

0.0095

0.01

0.0105

Baseline CogEasy CogHard VisEasy VisHard

Task

Me

an

dif

fere

nc

e o

f la

tera

l p

os

itio

n


Baseline --- 0.668 0.095 0.000 0.000

CogEasy --- 0.995

CogHard ---

VisEasy --- 1.000

VisHard ---


Volvo Technology

Quantifying loss of control (II): High-pass filtred standard deviation of lateral position

0.5

0.55

0.6

0.65

0.7

0.75

0.8


Task

SD

La

tera

l po

sit

ion

(h

igh

-p

as

s f

ilte

red

0.1

Hz)


Baseline --- 0.687 0.233 0.000 0.000

CogEasy --- 1.000 --- ---

CogHard --- --- ---

VisEasy --- 0.790

VisHard ---


Volvo Technology

Towards quantifying missed detection/interpretation Fraction consistent lane selected (any single lane selected

>75% of distance between two signs)

0.7

0.75

0.8

0.85

0.9

0.95


Task

Fra

cti

on

co

ns

iste

nt

lan

e

se

lec

ted


Baseline --- 0.735 0.060 0.000 0.000

CogEasy --- 0.972

CogHard ---

VisEasy --- 0.119

VisHard ---


Volvo Technology

Quantifying missed detection/interpretation: Fraction correct lane selected (of those consistently selected)


Baseline --- 0.000 0.000 0.000 0.000

CogEasy --- 0.000

CogHard ---

VisEasy --- 0.000

VisHard ---

0.8

0.85

0.9

0.95

1

1.05


Task

Fra

cti

on

co

rre

ct

lan

e s

ele

cte

d


Volvo Technology

Qualitative analysis of CogEasy vs. CogHard

0 500 1000 1500 2000 2500 3000

-5

0

5

Segment 1 (1 bl, 0 cog easy, 0 cog hard)

3000 3500 4000 4500 5000 5500 6000 6500

-10

0

10


6500 7000 7500 8000 8500 9000 9500

-10

0

10


0.95 1 1.05 1.1 1.15 1.2 1.25 1.3

x 104

-10

-5

0

5


1.3 1.35 1.4 1.45 1.5 1.55 1.6 1.65

x 104

-10

0

10


1.6 1.65 1.7 1.75 1.8 1.85 1.9 1.95

x 104

-10

0

10


1.95 2 2.05 2.1 2.15 2.2 2.25

x 104

-10

0

10


2.25 2.3 2.35 2.4 2.45 2.5 2.55 2.6

x 104

-10

0

10


2.6 2.65 2.7 2.75 2.8 2.85 2.9 2.95

x 104

-10

0

10


2.95 3 3.05 3.1 3.15 3.2 3.25

x 104

-10

0

10


Almost all errorsoccur for CogHard..


Volvo Technology

Conclusions

• Cognitive and visual tasks have different effects on lane changes• Cognitive tasks leads only to detection/interpretation errors – changing to wrong

lane or continuing straight ahead (mainly for CogHard)• Visual tasks mainly induce control errors (”overshoots”) but also some

detection/interpretation errors• These different effects clearly visible from qualitative analysis of lane

change plots• The effects can also be differentiated in quantitative analysis• Present quantitative results very preliminary – further analysis needed• Late detections not yet looked at (promising results from Trbovich)


Volvo Technology

How can these results be used in the LCT development?

• Provides a stronger scientific basis for the LCT – better understanding of why the deviation metric is so sensitive to both visual and cognitive load

• Consequences for calibration tasks – both visual and cognitive reference tasks needed (at least in theory)

• Specific ”control error” or ”detection error” metrics could be used to quantfy the visual and cognitive components (as complement to the standard LCT metric)

Volvo Technology Humans System Integration Volvo Technology Effects of visual and cognitive load on...

Documents

Transcript of Volvo Technology Humans System Integration Volvo Technology Effects of visual and cognitive load on...