News

Ego Vehicle factors that influence the passenger’s state in automated vehicles.

In one of the most complete and used model of comfort (Vick and Hellback, 2012), the authors suggested that the interaction of the user with the environment (person, product, usage) will produce subjective perception that, influenced by the user expectation, can result in comfort, discomfort or a neutral reaction. This model could be applied also to vehicle comfort, since the environment is composed by the vehicle characteristics, such as seat comfort (Elbanhawi et al., 2015), acoustic comfort (Zhang et al., 2018), thermal comfort (Danca et al., 2016), ride comfort (Cieslak et al., 2019), users state (psychological, physiological and physical), product usage (i.e. driving style) and the infrastructure and environmental characteristics (road condition, weather…).

However, the introduction of automated vehicle shifts the focus on passengers’ behaviour and perception. Factors, such as expected driving quality, motion sickness, communication, control and trust will either be introduced as new attributes of comfort or will increase their importance in the evaluation of passenger comfort. For instance, driver expectation will change focus point, as the passengers will not rely on a driver but they will rely on the car to behave in an acceptable and trustworthy way. Motion sickness will also have an impact on the user comfort: The absence of vehicle control (and therefore the impossibility to anticipate the trajectory of the car), the engagement in non-driving task and the new distribution of seats could increase the possibility to experience discomfort when being an automated vehicle user. In new connected autonomous vehicles, the passenger will also have to rely on the fact that the vehicle has information on the intention of another vehicle and this could affect the feeling of safety of the passenger. The capability of the automated vehicle to adapt to the state and perception of the passenger will be of outmost importance for the perception of comfort and safety.

Another dynamic factor which could influence the user comfort is the autonomous vehicle motion planning, which includes attributes like speed, acceleration, jerk profiles, condition for overtaking, headway distance and traffic law abiding. Various motion planning models have been developed in the literature. For example, the Model Predictive Control (MPC) (Alcalá et al., 2018) is proposed to improve comfort. The logic includes finding the future vehicle states using a dynamic model of the vehicle. This strategy was chosen to ensure that the vehicle path accomplishes specific motion requirements of a vehicle and that is achieved by considering lateral and longitudinal dynamics simultaneously. In other words, path and speed profile are defined at the same time. The MPC includes factors such as the boundary limits, the maximum speed, the sideslip rate, the longitudinal and lateral acceleration, jerks and time minimization and create an ideal path route. The advantage of route planning in automated vehicles is the possibility to modify the parameters described above depending on, for example, the state of the passenger.

Diels et al. (2017) developed an initial framework on the attributes that an automated vehicle, in this case Shared Autonomous Vehicles (SAV), should cover to enhance the users’ comfort.

Figure 1. Map of factors describing the concept of comfort in a Shared Autonomous Vehicle (SAV). Diels et al. (2017)

Considering this approach, in order to assess the overall vehicle comfort, ego vehicle comfort factors are investigated in SUaaVE by two different points of views, the dynamic comfort and the ambient comfort. The following sections describes SUaaVE approach and the main factors considered in dynamic and ambient comfort.

1. FACTORS OF DYNAMIC COMFORT

The attributes of dynamic comfort are divided into user experience attributes and vehicle dynamics attributes. The difference between the two groups rely more on the type of assessment they require. Specifically, user experience attributes can be assessed by non-expert assessors, while the vehicle dynamics attributes can only be assessed by dynamic experts. The figure below depicts the two groups of attributes and the factors included in these groups.

User experience high level attributes:

  • Ride Quality: Well-being of the vehicle’s occupants by controlling levels of accelerations and vibrations during vehicle travel.
  • Motion Sickness: Well-being of the vehicle’s occupants by controlling low frequencies body motion and absence of sickness.
  • Confidence: Perception of a safe and accurate operation of the vehicle given by the absence of corrections and non-expected behaviour.

Vehicle Dynamics specific attributes:

  • Precision: Capability to accurately follow a trajectory with minimum deviations of lateral error and yaw.
  • Stability: Capability to maintain a controllable trajectory that is perceived as safe.
  • Smoothness: Capability to perform trajectories within comfortable limits for vehicle states.
  • Disturbance sensitivity: Capability to maintain trajectory and isolate passenger feeling from unexpected external inputs like cross-wind and changes in the road grip.
  • Body control: Capability to control roll motion and pitch motion within comfortable limits.
  • Primary ride: Capability to control low frequency vertical and longitudinal motion and acceleration within comfortable limits.
  • Vibration isolation: Capability to isolate from sources of vibration and minimize accelerations.
  • Impact isolation: Capability to isolate passenger from deterministic road inputs.

2. FACTORS OF AMBIENT AND POSTURAL COMFORT

Regarding comfort, there could be identified seven main attributes:

  • Spatial Environment defines the level of perceived space within the occupant environment, with relation to the demands of the user (Lavieri and Bhat, 2019). A level of spatial comfort can be evaluated based upon the following factors:
  • Headroom (Proximity of vehicle trim to head).
    • Vertical:
      • Longitudinal.
      • Lateral.
    • Body room (Proximity of vehicle trim to body).
    • Degree of spatial separation to other occupants (Proxemics).
    • Cabin lighting characteristics.
    • Exterior visibility.
  • Thermal Environment covers the thermal environment of the vehicle cabin, following the individual characteristics and state of each user (Musat and Helerea, 2009). Thermal comfort evaluation is made up of the following components:
    • Ambient conditions meeting the occupant demands:
      • Ambient temperature.
      • Ambient humidity.
    • Airflow meeting demands of the occupant:
      • Speed.
      • Direction.
      • Temperature.
      • Humidity.
    • The capability for the cabin to adapt to a state which meets the context of use and the preferences of each occupant.
  • Acoustic Environment refers to components of the vehicle cabin sound and vibration characteristics (Nor et al., 2008). Acoustic comfort attributes are:
    • Sound levels within the vehicle cabin.
    • Tonality and frequency characteristics of sound.
    • The capability for the cabin to adapt to a state which meets the context of use and the preferences of each occupant.
  • Visual Environment refers to visible components of the vehicle occupant environment (Wienold, 2007). The measurement of visual comfort comprises of:
    • Lighting sources within cabin:
      • Natural light.
      • Artificial light.
    • Characteristics of light within the cabin:
      • Colour.
      • Intensity.
      • Source position.
      • Position of illumination.
    • Levels of discomfort glare.
    • The capability for the cabin to adapt to a state which meets the context of use and the preferences of each occupant.
  • Contact Surfaces (Tactile Interaction) concerns the comfort in interacting with systems requiring touching (e.g. touchscreen).
  • Postural position refers to components of the occupants’ physical position when traveling inside the vehicle cabin, following the characteristics and state of each occupant (Parida et al., 2018). The quantification of postural comfort is based upon the following aspects:
    • The vehicle seat providing a seated position which is compatible with the occupant activity:
  • Productive tasks.
  • Inter-occupant interaction.
  • Media consumption.
  • Relaxing / sleeping.
  • The capability for the cabin to adapt to a state which meets the context of use and the preferences of each occupant.
  • Provides acceptable degrees of transient posture.
  • Environmental Hygiene is comprised of a variety of factors governing the users sensation of cleanliness and hygiene whilst traveling within the vehicle (Lavieri and Bhat, 2019). The overall hygienic comfort of the vehicle comprises of factors including:
    • Air quality.
    • Cabin odor.
    • Cleanliness of surfaces.
    • Characteristics and behaviour of other users (shared use).
    • Vehicle state.
    • The capability for the cabin to adapt to a state which meets the context of use and the preferences of each occupant.

For more information about the ego vehicle factors that could influence the passenger’s state, please see Deliverable 3.1. Framework of the emphatic module and preliminary relationship among automotive factors with cognitive and emotional passenger state.

http://www.suaave.eu/wp-content/uploads/sites/17/2019/11/SUaaVE_WP3_D3.1_20191031_V100.pdf

News

External factors that influence the passenger’s state in automated vehicles

In an automated vehicle there are different external factors that could affect the passenger state, which are defined as completely external to the driver and the vehicle. A first approach to define and classify the external factors for the autonomous driving is found in Adaptive project (FP7-ICT-2013.6.5: Co-operative mobility Grant Agreement No. 610428) in D2.1. System Classification and Glossary.

The classification of the environment factors made by Adaptive is used as a basis for SUaaVE approach and for the development of ALFRED (a concept to humanise the behaviour of the vehicle according to the passenger state). Additionally, manoeuvre factors defined by Adaptive related with the interaction with other vehicles has been included.

Following Adaptive project, external factors aredividedin four categories: Traffic, Road, Visibility and Maneuvers.

1. TRAFFIC PARAMETERS

The main parameters to classify the traffic factors are:

.- Traffic participants (type of road users in the environment). These can be non-motorized road users and motorised road users.

.- Mix of participant according with automatization (defines whether the environment is shared with or without automated vehicles).

.- Traffic flow (which mixes the concepts of traffic speed and density).

  • Moving traffic: Traffic is moving nearly with recommended speed of particular road type. Traffic density is low or medium.
  • Slow moving traffic: Traffic is moving distinctly below recommended speed of particular road type. Traffic density is medium to high.
  • Stationary traffic: Traffic is nearly at a standstill or is at a standstill. Traffic density is high.

2. ROAD PARAMETERS

The main parameters to classify the road characteristics are:

  • Road condition. This factor describes the road according with smoothness and adhesion conditions (good, slippery and bumpy).
  • Road accessibility. This factor characterizes privacy of the road (public or private).
  • Road geometry. This factor synthesizes the geometry of the road in terms of curves and slopes (straight, curved or steep).
  • Road infrastructure. This synthesizes the different elements in the road.
    • Physical cut-off: Physical cut-off between oncoming lanes. Example: Guardrail, separating green area.
    • Good lane markings: White / yellow painted stripes or botts’ dots to separate lanes of a road.
    • Guard rails: Mechanical construction to prevent vehicles from veering off the roadway into oncoming traffic, crashing against solid objects or falling into a ravine. Examples: Guard rails, mural, concrete wall, taut steel rope, mound
    • Deer fences: Fence at the roadside which prevents animals and pedestrians from entering the road. Remark: “No deer fence” does not mean “no automation”. The evaluation of minimal infrastructure requirements for specific applications is a separate topic. Example: A Traffic Jam Pilot might not need a deer fence. For high speed application is has to be assessed if occurrence probability of deer in combination with perception performance results in an acceptable risk.
    • Emergency lanes: Separate lane at the roadside which is reserved for vehicles with technical defects. Remark: Hard shoulders is a synonym for emergency lane
    • Traffic light: Traffic light at intersections of e.g. urban or rural roads.
    • Road type. Main road types are: Motorway, Highway, Interstate, Rural road, Arterial road, Urban road, Residential district roads, Parking area & parking deck and Garage.

3. VISIBILITY DUE TO EXTERNAL CONDITIONS PARAMETERS

Visibility factors are included in SUaaVE approach to define easiness of visibility of environment by the passengers inside the car due to the external conditions. These are:

.- Good visibility: Full visibility of vehicles and obstacles. Remark: Modest fog, spray, rain or snow shall not hamper system functionality.

.- Poor visibility due to obstacles (vehicles or infrastructure).

.- Poor visibility due to weather conditions (fog, heavy spray, heavy rain, heavy snow).

4. EXTERNAL PARAMETERS OF MANEUVER (RELATED WITH OTHER VEHICLES)

In Adaptive, classification vehicle maneuver is characterized by the following parameters: maneuver time to collision, maneuver duration, maneuver automation, maneuver speed range, maneuver control force, maneuver time headway, maneuver trigger and maneuver coordination.

For external factors characterization, SUaaVE considers only the parameters related with maneuver in relation with the interaction with other vehicles in the road, i.e. maneuver time to collision, maneuver time headway and maneuver coordination. Maneuver duration, maneuver automation, maneuver speed range, maneuver control force and maneuver trigger are included in dynamic comfort parameters (Ego Vehicle factors).

  • Vehicle manoeuvre time-to-collision. This factor characterizes if the collision is imminent or not.
    • Large: Collision is not imminent. Example: Driver assistance systems such as ACC, LKA, etc.
    • Small: Collision is imminent. Example: emergency braking e.g. if lead vehicle brakes hard suddenly.
  • Vehicle manoeuvre time headway: This factor characterizes the in seconds the distances between vehicles.
    • Standard: Time headway > 0,9 sec Examples: ACC, Traffic Jam Assistance.
    • Reduced: Time headway 0,5 … 0,9 sec Example: truck platooning with 15m distance.
    • Small: Time headway < 0,5 sec Example: truck platooning with 5m distance.
  • Vehicle manoeuvre coordination: This factor characterizes the coordination nature of the manoeuvre.
    • With coordination: Manoeuvre involves several vehicles which are coordinating their behaviour. Example: Automated filtering at on-ramp of a motorway – vehicle that wants to enter motorway asks vehicles on motorway via V2V communication to increase headway so to ease filter-in manoeuvre.
    • Without coordination: Manoeuvre is not coordinated between involved vehicles. Example: Lane change at overtaking manoeuvre – if the adjacent lane is not occupied the lane change is initiated without any coordination or communication between involved vehicles.

For more information about the external factors that could influence the passenger’s state, please see Deliverable 3.1. Framework of the emphatic module and preliminary relationship among automotive factors with cognitive and emotional passenger state.

http://www.suaave.eu/wp-content/uploads/sites/17/2019/11/SUaaVE_WP3_D3.1_20191031_V100.pdf

News

Predicting Acceptability of Connected Automated Vehicles

Connected Automated Vehicles (CAV) are completely self-driving cars that are equipped with tools to communicate and share data with other devices both inside and outside the car, such as other cars, and public transport systems. The driver has the role of a passenger inside CAV. CAV is not available on the market right now, but they may dominate the road environment in the future. Many car manufacturers have already incorporated some degrees of automation into their cars, such as parking assist, and adaptive cruise control. Some manufacturers are now pilot testing vehicles with high or full automation in designated areas.

CAV may play an important role to solve several societal problems, such as increasing traffic safety, reducing traffic jams, enhancing mobility for those unable to drive, and reducing traffic CO2 emissions.

SUaaVE Project

In the SUaaVE project (Horizons 2020 project funded by the EU) we examine the acceptability and acceptance of CAV. Acceptability is an attitude people have towards CAV before they have experienced it. Acceptance is related to if people want to use or buy CAV after they have experienced it. We want to find out what drives the acceptability and acceptance of CAV, and what we can do to increase acceptability and acceptance for both potential users, as well as for other road users. SUaaVE aims to make a change in the current situation of public acceptance of CAV by focusing on the human side to improve more “intangible” aspects as safety perception, attitudes, and in general, emotional appraisal of CAV. First, we have examined what factors predict the acceptability of CAV. Based on an extensive literature review and several focus groups conducted in Italy, Spain, France, and the Netherlands, we proposed a psychological model that can predict acceptability of CAV.

Large Scale Survey in 6 European Countries

To test our psychological model that predicts the acceptability of CAV, we conducted a large scale survey. The survey was conducted online in 6 European countries: the United Kingdom, the Netherlands, Germany, France, Spain, and Italy. In total, almost 3800 people filled out the survey in April 2020. The sample was relatively evenly spread in terms of age (about 20% was between 18 and 30 years old, and about 20% was older than 55), gender, and country (about 630 participants per country). In the survey we explained what CAV is, and measured acceptability, how they perceived different aspects of CAV, and also measured various individual differences such as interest in technology and personal values.

Predicting Acceptability of CAV

We found that acceptability of CAV is predicted by its attributes, the perceived adoption norm, and the perceived behavioral control. The perceived adoption norm is the extent to which someone believes close others (such as friends, family, and coworkers) will adopt CAV in the future. The perceived behavioral control is the extent to which someone believes they will be capable of using CAV. Both of these positively predicted acceptability of CAV. As for CAV’s attributes, we found 7 distinct characteristics of CAV that influence acceptability: perceived safety (is CAV safe?), perceived convenience (is CAV useful?), perceived control (can I control CAV’s behavior?), perceived pleasure (is driving CAV enjoyable?), trust in CAV technology (is CAV’s computer system trustworthy?), perceived environmental sustainability (is CAV environmentally friendly?), and perceived status-enhancement (is CAV a status product?). Out of these, perceived safety, perceived convenience, and perceived environmental sustainability had the strongest positive effects on acceptability of CAV. Based on these results, we formulated some initial guidelines to enhance public acceptability.

More Information

Aside from the direct predictors of acceptability, we have examined other factors that could influence acceptability of CAV. For example, we have examined differences between drivers and non-drivers, and effects of personal values, the need for control, and experience with car technology. The full results are available as open access on www.suaave.eu/results. On this website you can read more about the SUaaVE project in general, as well as read all currently published open access deliverables.

News

Internal factors that influence the passenger’s state in automated vehicles

There are factors concerning the trip and the driver that are crucial when describing the emotional and cognitive state of the passenger. This can be divided into two categories: Profile-passenger factors and trip-passenger factors.

1. PROFILE-PASSENGER FACTORS

The characteristics of the passenger could affect their perception of the driving experience. As a first approach, the category “Profile-Passenger factors” includes these factors:

  • Gender.
  • Age.
  • Origin.
  • Culture.
  • Experience in driving.
  • Driving predisposition.
  • Personality.
  • Past experiences.
  • Physical and cognitive characteristics.

2. TRIP-PASSENGER FACTORS

In addition to the elements that constitute the environment of the trip, there are some trip characteristics that are also of great relevance for the cognitive and emotional description of the passenger. Those characteristics are the ones related with the link between the passenger and the trajectory. These can be classified as follows:

  • Time to complete the trip: From ample time, to very short time, to irrelevant.
  • Importance of get in time: From very important, to irrelevant.
  • Purpose of the trip: Work, hospitalization, holidays, etc.
  • Pre-journey and post-journey activity.
  • Familiarity with the trip // with the environment: prompt, regular // very familiar, completely new.

As an example of the previous factors, the driving experience and how the passenger perceives it, can be crucially different according to the purpose of its trip. For instance, if the purpose of the trip is holidays, the dense traffic could not get on his/her nerves in the same way, as if the trip is to get to work being short in time.

During the development of the emotional and cognitive models in SUaaVE, these factors will be identified and defined according with their relevance regarding the passenger’s state. For more information about other factors that can influence the passenger’s state, please see Deliverable 3.1. Framework of the emphatic module and preliminary relationship among automotive factors with cognitive and emotional passenger state.

http://www.suaave.eu/wp-content/uploads/sites/17/2019/11/SUaaVE_WP3_D3.1_20191031_V100.pdf

News

Special Session of SUaaVE at CHIRA 20 Conference

The Special Session on “Reliable Estimation of Passenger Emotional State in Autonomous Vehicles (SUaaVE 2020)” was held online November 6th at the International Conference on Computer-Human Interaction Research and Applications 2020 (CHIRA).

In this session, Lucie Lévêque (from Université Gustave Eiffel -IFSTTAR in the past-, one of the SUaaVE’s partners) presented the paper “Development of an Immersive Simulation Platform to Study Interactions between Automated Vehicles and Pedestrians”. The paper focuses on a novel simulation platform, the V-HCD, allowing the conduct of immersive experimentations, both from the pedestrian’s and the driver’s point of view. This platform will be used to study the acceptance of the automated vehicle in SUaaVE, and further to support the human-centred design of a future empathic automated vehicle (AV).

Figure 1: Example of scenario implemented on the V-HCD platform to study the interaction between a pedestrian and an AV with a more or less attentive driver. Université Gustave Eiffel ©.

Juan Felipe Medina-Lee (from the Center for Automation and Robotics, researcher in the project ‘Programmable systems for intelligence in automobiles’ as a member of Autopia program) presented “Traded Control Architecture for Automated Vehicles Enabled by the Scene Complexity Estimation”. The research consists on a novel traded control architecture proposed to enhance the operational domain of the autonomous driving system (ADS) under the premise that vehicles and humans may need to adapt their cooperation level depending on the context.

The third presentation was conducted by Juan-Manuel Belda-Lois (researcher at Instituto de Biomcecánica de Valencia, IBV, the coordinator partner of the SUaaVE project) with the title “The Estimation of Occupants’ Emotions in Connected and Automated Vehicles”. This research covers an initial experiment to identify changes in the emotional state of the occupants in different driving experiences (in a driving simulator and in real conditions) by measuring and analysing the physiological signals of the participants, serving as a basis for the generation of the emotional model. The results showed that it is possible to estimate the level of Arousal and Valence of the participants during the journey from the analysis of ECG, EMG and GSR signals.

Figure 2: HRV along the experiment and emotional components (Arousal and Valence) declared by a co-driver in each event of an urban journey by car. IBV ©.

The last presentation of the session was conducted by Davide Salanitri (from IDIADA, partner of the SUaaVE project) with the title “Evaluation of a New System in Future L4 Vehicles: Use Cases and Methodology for the SUaaVE European Project”. The paper describes the definition process of the use cases in SUaaVE, considering the different factors that characterize them.

Finally, in the session “Interaction Design” at CHIRA 2020, Benjamin Chateau (from CATIE, another SUaaVE’s partner) presented the paper “Exploring Empathetic and Cognitive Interfaces for Autonomous Vehicles”. An interface for automated vehicles capable of informing the user at any time about the road situation and reassuring him/her about the information processed by the vehicle.

Figure 3: Interface overview. CATIE ©.