Human Factors Studies for Limited- Ability Autonomous Driving Systems (LAADS)

Transcription

1 Human Factors Studies for Limited- Ability Autonomous Driving Systems (LAADS) Glenn Widmann; Delphi Automotive Systems Jeremy Salinger; General Motors Robert Dufour; Delphi Automotive Systems Charles Green; General Motors Eddy Llaneras; Virginia Tech Transportation Institute 1

2 LAADS Assumptions & Definitions LAADS Definition Lane Centering is only available with Adaptive Cruise Control Together they are the Limited Ability Autonomous Driving System (LAADS) LAADS may operate for extended periods without the driver touching the steering wheel. Active Safety sensing performance has limitations. Driver must be responsible for vehicle operation, including headway, lane positioning, object and event detection, and avoidance. Central Value Proposition of LAADS Enables hands-free driving activities potentially increased comfort and reduced manual workload Key Human Factors Issues Assisting driver to maintain responsibility while LAADS is performing primary control help drivers avoid: a. becoming over-reliant upon the systems, b. operating the systems outside of design parameters, or c. not being aware when the systems are not operating as intended. 2

3 Research Phases & Environments Research Phases 1. Problem Identification Research Determine driver behaviors & secondary tasks that naturally occur with LAADS 2. Alternative Driver-Vehicle Interaction Concept Research Formative Usability evaluations to down select and refine countermeasure detailed interaction Pilot quantitative experimental studies to measure countermeasure effectiveness 3. Integrated Countermeasure Concept Research Quantitative experimental studies to measure countermeasure effectiveness Study Environments Surveys, Expert Panel Studies, Driving Simulator Studies, Track Studies 3

4 LAADS HW/SW Environment in Support of Human Factors Research Goals DriveSafety DS-600c Lane Centering Driver Monitor System headway-gap 2009 Chevrolet Malibu Adaptive Cruise Control 4

5 Human Machine Interface Description Driver seat Elbow switches Haptic seat (vibration) Steering wheel left/right touch sensor Lightbar Foot Switch left/right selection Exogenous Display LCD Display Telltales Lightbar Driver State Monitor Touch Sensing ACC & LC Switches Elbow Switches LCD Display Foot Switch Haptic Seat 5

6 Phase 1: Simulator Environment Design Drivers instructed purpose of study was to evaluate ACC & LAADS system, NOT examine their behavior System options ACC Driving (ACC-only) Limited Ability Autonomous Driving System (LAADS) Perfect Autonomous Driving System (PADS) Long drive (~2 hours) Secondary tasks readily available (affordances) Food & drink; IPOD with music library; Reading material; Personal cell phone System Limitation events: Poor lane markings; Decelerating lead vehicle; Excessive curve; Lane drifts; Construction Secondary Tasks are Naturalistic Occurrences on the part of Drivers Experimenter leaves lab and not be available until end of drive. 6 6

7 Secondary Tasks Engagement Type of interactions Drivers interacted with many objects Drinking & eating Car radio/cd player & cds; MP3/IPod Mobile phone Video-viewing & Internet access also observed ACC PADS LAADS Fatigue also observed Mean Frequency of Interaction (# of occurrences per test drive) Music (radio, ipod, CD's) Cell phone total 7 7

8 Driver Monitoring Head Turn Frequency by Duration Category Frequency of Eccentric Head Turns by duration categories Eccentric = Any head turn beyond boundary 8 8

9 Phase I: Test Track Environment Design Drivers instructed purpose of study was to evaluate ACC & LAADS system, NOT examine their behavior Long drive (~3 hours) Secondary tasks readily available Food & drink IPOD with music library Reading material Personal cell phone Secondary Tasks are Naturalistic Occurrences on the part of Drivers Experimenter accompanies driver, but checks out doing personal work Newspaper Reaching 9

10 Riskier Tasks Tended to be Limited to LAADS Driving (those with relative risk values above 1) Percentage of Drivers Observed to Engage 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 67% 83% 17% 92% 92% 92% 75% 42% 58% 8% 83% 17% 67% 25% 58% 8% 8% 42% ACC Driving LAADS Driving 67% 50% 42% 25% 8% 25% 8% 50% 0% 0% 0% 0% 0% Relative risk values based on 100-Car Study (value of 1 is crash risk normally associated with typical driving) 10

11 LAADS Also Found to Increase the Percentage of Unsafe Glances (Those Extending Beyond 2 Seconds) 120% 110% 100% 112% 113% 90% 80% 74% 70% 60% 54% 50% 40% 40% 41% 30% 27% 20% 10% 0% 0% 0% 0% 0% 0% 0% Overall

12 Phase II Formative Research Purpose Select candidate Human Machine Interface (HMI) subsets for further testing Telltales (icons and visual indicators) Graphic display arrangement, ordering and timing Driver I/O (switches and buttons) Formative Study Survey study Structured method to select starting HMI candidates Usability study Have drivers provide feedback as to combinations of features in the context of driving 12

13 Summary Of Research Results Phase I (Sim & Vehicle tests) No Countermeasures Studies identified the need for interaction countermeasures under LAADS driving. LAADS found (when compared to no-laads driving) to Change the propensity of secondary task engagements. Secondary task engagements generally increase under LAADS driving. Increase riskier behaviors (those requiring extended off-road glances) such as reading, reaching for object in back seat, texting. Negatively impact the degree of visual attention drivers devote to the forward roadway increases off-road glances Phase II (Sim & Vehicle tests) Studies generated a set of interaction concepts designed to overcome or counter visual inattention problems identified above (in Phase I) Includes Convenience Messages, System State Messages (Limitations, and Faults), and Visual Inattention Countermeasures Current data suggest that countermeasures are understood by drivers and appear to be effective Phase III will evaluate the set 13

14 Human Factors for Limited-Ability Autonomous Systems Team Roster Vehicle Manufacturer Tier 1 Automotive Suppliers Universities Systems and Safety Engineering Program partially funded by Federal Highway Administration 14

15 Thank You & Questions?? 15