Institute for Transport StudiesFaculty of Environment

What do psychology and neuroscience have to do with vehicle engineering? – The driver model case for cross-disciplinarity

Dr. Gustav MarkkulaHuman Factors & Safety GroupITS Research Seminar2016-11-17

Vehicle engineering?

The driver

Driver models

”The virtual crash test dummy with a brain”

Conflicting descriptions?

Routine driving Near-crash driving

Closed-loop Open-loop

Short delays Long, random delays

Well-adjusted control Under- and overreactions

(Van Auken et al., 2011)

(MacAdam et al., 2003)

Stealing ideas

Motor primitives

(Flash and Henis, 1991)

(Cook and Maunsell, 2002)Evidence accumulation

Perceptual heuristics

(Land and Horwood, 1995; Wann and Wilkie, 2004; Salvucci and Gray, 2004)

Routine driving Near-crash driving

Closed-loop Open-loop

Short delays Long, random delays

Well-adjusted control Under- and overreactions

Conflicting descriptions?

Intermittent open-

loop adjustments...

... timing from

evidence accumulation

... amplitudes from

perceptual heuristics,

near-optimal for routine

circumstances

(Mar

kkul

a, 2

014)

Steering during skidding

�̇� ( 𝑡+𝑇 R )=𝑘f �̇� f (𝑡 )≈−𝑘𝑓 �̇�(𝑡 )

(Markkula, 2013; Markkula, Benderius, Wahde, & Wolff, 2014)

Understanding near-crash brake response

(Markkula, 2014)

-1 -0.5

-1

-0.5

0

0.5 Truck crash

R2 = -0.08

0 50

0.5

1

-1 -0.5 0 0.5

-1

-0.5

0

0.5 Bus near-crash

R2 = 0.98

0 50

0.5

1

-3 -2 -1 0

-1

-0.5

0

0.5 SHRP 2near-crash

R2 = 0.18

0 50

0.5

1

-6 -4 -2

-1

-0.5

0

0.5 SHRP 2crash

R2 = 0.60

0 50

0.5

1

-1 -0.5

-1

-0.5

0

0.5 ANNEXTcrash

R2 = 1.00

0 50

0.5

1

-3 -2 -1 0

-1

-0.5

0

0.5 SHRP 2near-crash

R2 = 0.75

0 50

0.5

1

-2 -1

-1

-0.5

0

0.5 SHRP 2crash

R2 = 0.98

0 50

0.5

1

-4 -2 0

-1

-0.5

0

0.5 SHRP 2near-crash

R2 = 0.87

0 50

0.5

1

-4 -2

-1

-0.5

0

0.5 SHRP 2crash

R2 = 0.97

0 50

0.5

1

-4 -2

-1

-0.5

0

0.5 Truck crash

R2 = 0.92

0 50

0.5

1

-6 -4 -2

-1

-0.5

0

0.5 Truck crash

R2 = 0.91

0 50

0.5

1

-1 -0.5

-1

-0.5

0

0.5 Truck crash

R2 = -0.08

0 50

0.5

1

-1 -0.5 0 0.5

-1

-0.5

0

0.5 Bus near-crash

R2 = 0.98

0 50

0.5

1

-3 -2 -1 0

-1

-0.5

0

0.5 SHRP 2near-crash

R2 = 0.18

0 50

0.5

1

-6 -4 -2

-1

-0.5

0

0.5 SHRP 2crash

R2 = 0.60

0 50

0.5

1

-1 -0.5

-1

-0.5

0

0.5 ANNEXTcrash

R2 = 1.00

0 50

0.5

1

-3 -2 -1 0

-1

-0.5

0

0.5 SHRP 2near-crash

R2 = 0.75

0 50

0.5

1

-2 -1

-1

-0.5

0

0.5 SHRP 2crash

R2 = 0.98

0 50

0.5

1

-4 -2 0

-1

-0.5

0

0.5 SHRP 2near-crash

R2 = 0.87

0 50

0.5

1

-4 -2

-1

-0.5

0

0.5 SHRP 2crash

R2 = 0.97

0 50

0.5

1

-4 -2

-1

-0.5

0

0.5 Truck crash

R2 = 0.92

0 50

0.5

1

-6 -4 -2

-1

-0.5

0

0.5 Truck crash

R2 = 0.91

0 50

0.5

1

Predictions for (near-/)crashes:• Piecewise near-linear deceleration• Above threshold, brake response

time decreases with urgency• Deceleration ramp-up slow, but

increasing with urgency

(Markkula, Engström, Lodin, Bärgman & Victor, 2016)

Time (s)A

ccel

erat

ion

(g)

Time (s)

Explaining variability in near-crash response time

(poster by Markkula, Lodin, Wells, Theander & Sandin, 2016)

(a number of slides with unpublished work removed for web sharing)

Related application: Transitions from automated to manual driving

(Merat, Jamson, Lai, Daly & Carsten, 2014)

(Zeeb, Buchner & Schrauf, 2015)

(a number of slides with unpublished work removed for web sharing)

Summary / reflections

• Reaching from an applied question to ideas and concepts from more basic sciences

• Key here: Constrained applied problem provided focus

• Resulting cross-disciplinary connection proved a rich source of new ideas and applications

• Mechanisms grounded in more basic sciences applied models more likely to generalise

Pros and cons of cross-disciplinarity

• Potential for low-hanging fruit• Personal development, fun

• More difficult to communicate findings?

Institute for Transport StudiesFaculty of Environment

Thanks!