External HMI of communication and autonomous vehicles: a ...€¦ · light No pedestrian crossing A...
Transcript of External HMI of communication and autonomous vehicles: a ...€¦ · light No pedestrian crossing A...
#SMIV2019 Annual Scientific Conference
External HMI of communication and autonomous vehicles: a pedestrian’s study
Métayer, N., Bonneviot, F., Cherni, H., Coeugnet, S., & Souliman, N.
VEDECOM mobiLAB - 23 bis allée des Marronniers – 78000 Versailles - FRANCE
2
#SMIV2019 Annual Scientific Conference
SUMMARY
1. The project « Paris Saclay Autonomous Lab »
2. Challenges
3. Theoritical elements
4. Study
5. Results
6. Conclusion
3
#SMIV2019 Annual Scientific Conference
THE PROJECT « PARIS SACLAY AUTONOMOUS LAB » (EVAPS)
4
#SMIV2019 Annual Scientific Conference
OBJECTIVES
Objective: develop Disruptive Intelligent Mobility services to cover peri-urban
circuits:
- On Paris-Saclay territory
- With autonomous driving (= without driver)
- On private site and dedicated lanes
- With two types of electrical vehicles (car and shuttle)
- With the last kilometer included
5
#SMIV2019 Annual Scientific Conference
FUNDING INSTITUTIONS & PARTNERS
6
#SMIV2019 Annual Scientific Conference
CHALLENGES
7
#SMIV2019 Annual Scientific Conference
CHALLENGES
Pedestrians and the eHMI
- Around the AV, there is not only the driver but also a lot of people ;
- A lot of researches on drivers, less on pedestrians ;
- With the AV, a human is not necessarily in the vehicle so there is no more non-verbal communication between the driver and the pedestrians ;
→ Question: Is an eHMI necessary to communicate the vehicle intention to the pedestrians?
- VEDECOM is involved in the ISO group for this question.
8
#SMIV2019 Annual Scientific Conference
THEORITICAL ELEMENTS
9
#SMIV2019 Annual Scientific Conference
INTRODUCTION
- Current research seems to show the importance of an eHMI (e.g., de Clercq et al., 2019; Lagström &
Lundgren, 2015; Mahadevan et al., 2018b, 2018a; Matthews et al., 2017; Schieben et al., 2019)
- Many eHMIs exist and are tested
- Display (Ackermann et al., 2019; Clamann et al., 2017; de Clercq et al., 2019)
- Band of LEDs (Ackermann et al., 2019; Faas & Baumann, 2020; Gruenefeld et al., 2019; Lagström & Lundgren, 2015)
- Eyes (Chang et al., 2017; de Miguel et al., 2019; Mahadevan et al., 2018b, 2018a)
- Hands (Mahadevan et al., 2018b, 2018a)
Clamann et al., 2017Ackermann et al., 2019Chang et al., 2017Mahadevan et al., 2018
10
#SMIV2019 Annual Scientific Conference
INTRODUCTION
- Pedestrians do not necessarily perceive eHMIs (Chang et al., 2017)
- Pedestrians consider the distance between them and vehicles and their speed when crossing (Clamann et al., 2017; Dey et al., 2017, 2019)
- Pedestrians correctly manage their interactions with the AV without eHMI(Rothenbücher et al., 2016)
- Without an eHMI, only 13% of pedestrians cross in front of an AV before it was completely stopped, compared to 38% for an AV with an eHMI(Lagström & Lundgren, 2015; Lee et al., 2019)
- Presence of a pedestrian crossing have an impact on pedestrian’s behaviour (e.g., Clamann et al., 2017; Jayaraman et al., 2018)
11
#SMIV2019 Annual Scientific Conference
HYPOTHESIS
12
#SMIV2019 Annual Scientific Conference
HYPOTHESIS
H1: Pedestrians would cross the street more often in front of a vehicle with an eHMI than a vehicle without eHMI… (e.g., de Clercq et al., 2019; Schieben et al., 2019)
H2: … especially when there is no pedestrian crossing (e.g., Clamann et al., 2017;
Jayaraman et al., 2018).
H3: No objective difference would be observed between eHMIs.
H4: The subjective data should highlight a preference for one system rather than the two others (e.g., Ackermann et al., 2019).
13
#SMIV2019 Annual Scientific Conference
STUDY
14
#SMIV2019 Annual Scientific Conference
METHOD
• Participants• 49 participants (24 men and 25 women, Mean age = 41.02 years old, SD = 12.3)
• Material• Many questionnaires
• The street-crossing assessment questionnaire
• Acceptance scale (Van Der Laan, Heino, & De Waard, 1997)
• Presence questionnaire (Witmer & Singer, 1998)
• Preference
• Learning stage of the eHMI
• Virtual environment
• Urban environment
• 5 buildings
• 5 crossings
15
#SMIV2019 Annual Scientific Conference
2 Start1
2
3 4
51
3 4
A top view of virtual environment
16
#SMIV2019 Annual Scientific Conference
TWO INDEPENDENT VARIABLES
- Road infrastructure
- Vehicle
17
#SMIV2019 Annual Scientific Conference
ROAD INFRASTRUCTURE
A pedestrian crossing without pedestrian traffic
light
No pedestrian crossing
A pedestrian crossing with pedestrian traffic
light
18
#SMIV2019 Annual Scientific Conference
VEHICLES
Conventional = with a driver
Autonomousvehicle with eHMI
LED light strips
Autonomousvehicle
Pictograms
Threads of diodes
19
#SMIV2019 Annual Scientific Conference
THE EXTERNAL HMI
4 messages :
- The vehicle circulates
- The vehicle brakes
- The vehicle is stopped
- The vehicle starts
20
#SMIV2019 Annual Scientific Conference
MAIN QUESTION
What is the pedestrian’s behaviour when hehas to cross ?
21
#SMIV2019 Annual Scientific Conference
EXAMPLES
One configuration :
➢ The autonomous vehicle without eHMI and no pedestrian crossing
➢ All vehicle stop when they detect a pedestrian even if there is no pedestrian crossing
Does the pedestrian pass or not?
22
#SMIV2019 Annual Scientific Conference
PROTOCOLE
1 week before
study
Day of study
3 times
23
#SMIV2019 Annual Scientific Conference
RESULTS
- The crossing behaviour
- The subjective measures
24
#SMIV2019 Annual Scientific Conference
The crossing behaviour
25
#SMIV2019 Annual Scientific Conference
GO OR NO GO ?Vehicle
No go Go
Thread of diodes 1% 99%
LED light strips 1% 99%
Pictograms 2% 98%
Autonomous vehicle without eHMI 8% 92%
Conventional vehicle 5% 95%
Significant difference between vehiclewith eHMI and vehicle without eHMI
26
#SMIV2019 Annual Scientific Conference
GO OR NO GO ?Road structure
No go Go
Without pedestrian crossing 8% 92%
With pedestrian crossing 2% 98%
With pedestrian crossing and lights 0% 100%
Significant difference between no protectedcrossing and protected crossing
27
#SMIV2019 Annual Scientific Conference
And the combination of the two variables ?
28
#SMIV2019 Annual Scientific Conference
GO OR NOT GO ?
Vehicle Road Structure No go Go
Without eHMI
No pedestrian crossing 14% 86%
With pedestrian crossing 2% 98%
With eHMI
No pedestrian crossing 4% 96%
With pedestrian crossing 0% 100%
29
#SMIV2019 Annual Scientific Conference
The subjective measures
30
#SMIV2019 Annual Scientific Conference
eHMIS’ UTILITY AND SATISFACTION
-2
-1,5
-1
-0,5
0
0,5
1
1,5
2
Utility Satisfaction
Threads of diodes LED light strips Pictograms
No difference
31
#SMIV2019 Annual Scientific Conference
PREFERENCE
6%
49%
45%
1st rank
Threads of diodes LED light strips Pictograms
*
*
The threads of diodes were much less selected as first compared to the other two eHMIs.
32
#SMIV2019 Annual Scientific Conference
PREFERENCE – MESSAGE BY MESSAGE
The vehiclecirculates
The vehicle brakesThe vehicle is
stoppedThe vehicle starts
1st rank
3rd rank
33
#SMIV2019 Annual Scientific Conference
PREFERENCE – MESSAGE BY MESSAGE
The vehiclecirculates
The vehicle brakesThe vehicle is
stoppedThe vehicle starts
1st rank
LED light strips LED light strips LED light strips LED light strips
Pictograms Pictograms
3rd rank
34
#SMIV2019 Annual Scientific Conference
PREFERENCE – MESSAGE BY MESSAGE
The vehiclecirculates
The vehicle brakesThe vehicle is
stoppedThe vehicle starts
1st rank
LED light strips LED light strips LED light strips LED light strips
Pictograms Pictograms
3rd rank
Threads of diodes Threads of diodes Threads of diodes Threads of diodes
Pictograms Pictograms
35
#SMIV2019 Annual Scientific Conference
Why these choices?
36
#SMIV2019 Annual Scientific Conference
THE MAIN REASONS
1st rank 3rd rank
37
#SMIV2019 Annual Scientific Conference
THE MAIN REASONS
1st rank 3rd rank
The visibility
The ease of understanding
38
#SMIV2019 Annual Scientific Conference
THE MAIN REASONS
1st rank 3rd rank
The visibility The lack of visibility
The ease of understanding The complexity of messages
39
#SMIV2019 Annual Scientific Conference
CONCLUSION
40
#SMIV2019 Annual Scientific Conference
HYPOTHESIS RECALL
Hypothesis Results
41
#SMIV2019 Annual Scientific Conference
HYPOTHESIS RECALL
Hypothesis Results
H1: Pedestrians would cross the street more often in front of a vehicle with an eHMI than a vehicle without eHMI…
42
#SMIV2019 Annual Scientific Conference
HYPOTHESIS RECALL
Hypothesis Results
H1: Pedestrians would cross the street more often in front of a vehicle with an eHMI than a vehicle without eHMI…
H2: … especially when there is no pedestrian crossing.
43
#SMIV2019 Annual Scientific Conference
HYPOTHESIS RECALL
Hypothesis Results
H1: Pedestrians would cross the street more often in front of a vehicle with an eHMI than a vehicle without eHMI…
H2: … especially when there is no pedestrian crossing.
H3: No objective difference would be observed between eHMIs.
44
#SMIV2019 Annual Scientific Conference
HYPOTHESIS RECALL
Hypothesis Results
H1: Pedestrians would cross the street more often in front of a vehicle with an eHMI than a vehicle without eHMI…
H2: … especially when there is no pedestrian crossing.
H3: No objective difference would be observed between eHMIs.
H4: The subjective data should highlight a preference for one system rather than the two others.
45
#SMIV2019 Annual Scientific Conference
CONCLUSION
- The presence of eHMI seems important especially when there is no
protected crossing
- No behaviour difference appears between the 3 eHMIs
- The participants do not like the threads of diodes eHMI (complexity
and lack of visibility), they prefer pictograms and LED light strips
(visibility and ease of understanding)
→ What’s next?
Develop a prototype of eHMI and test it on the road