Presentation on theme: "A FCW, CIB and DBS benefit estimation method developed in ACAT program"— Presentation transcript:
1A FCW, CIB and DBS benefit estimation method developed in ACAT program Hirofumi Aoki, Masami Aga, Yoshiki Miichi, Yoshiaki Matsuo, and Shin TanakaToyota Motor CorporationThank you Mr. Fanke for introducing me.I’m very grad to be here.And I’m also very grad to livenear Washington DC for some years,for I have many good memories in the United States.Our company has an unpleasant memory though.Today I would like to talk about“A FCW, CIB and DBS benefit estimation method”.The target system is mainly designed to help reduceor mitigate rear-end collisions.The estimation method had been developed a few years ago,and we made some modifications applying it to our systems,the concept has not changed.I would like to present youthe essence of our method.
2Developing Systems to Reduce/Mitigate Rear-end Collisions SubaruToyotaMercedesHondaAs you all well aware,various systems have appeared in the marketover the past several years.The sensors to detect proceeding vehiclesvaryfrom system to system,and their control strategy also varyfrom system to system.Volvo
3CIB (Automated braking) Control SequenceOrdinary DrivingCollisionTimeSpeedDBS (Brake Assist)FCWCIB (Automated braking)The full system has 3 functions shown here.They areFCW or collision warning,DBS or brake assist,CIB or automated braking.[Click]They work togetherto provide collision mitigationor collision avoidance.Toyota’s objective in ACAT was to investigatehow these 3 functions affect on the system benefit.However, it is not easy.The objective in ACATFCW, DBS and CIBspecificationSystembenefit
4Drivers reaction to Benefit Ordinary DrivingCollisionFCW+ DBSCIBDBS (Brake Assist)FCWCIB (Automated braking)TimeSpeedImagine that the drivers show higher response to the warning,[Click]then FCW & DBS are beneficial and CIB’s role is small.
5Drivers reaction to Benefit Ordinary DrivingCollisionFCW+ DBSCIBDBS (Brake Assist)FCWCIB (Automated braking)TimeSpeedImagine that the drivers show poorer response to the warning,[Click]then FCW & DBS are less effective and CIB’s role is larger.It means that (Thus),Each benefit significantly depends on“how drivers react before collisions”.So,So…Each benefit significantly depends on“how drivers react before collisions”.
6Driver behavior investigation using the Driving SimulatorTo measure such driver responses,Toyota created rear-end collision scenarios in the driving simulator.More than 100 ordinary drivers participated in the test.
7Driving Simulator Test Example Here, I will show you a driving simulator test example.[Video]Here, the warning was issued just before a collision.The collision scenario was given just once for each participant.After the tests, most of the drivers saidthat was their first rear-end collision experienceand they were very astonished.We consider that drivers act in the same mannerin actual rear-end collisions.7
8Driver Reaction Before Collisions [m/s2]Maximumdecelerationss60s -10(31 persons in total)8642This chart shows the result.Here, the X-axis indicatesthe period from warning to braking onset,and the Y-axis indicates the maximum deceleration.It indicatesthat the driver reaction varies from person to person,And there’s poor relation betweenthe period from warning to braking onsetand the maximum deceleration.It meansthat both variables are considered to be independent.0.01.02.0[s]Period from warning to braking onset
9Rear-end Reconstruction Concept How are rear-end collisions happening?Traffic accidents are happening incidentallyThen, how are rear-end collisions happening in the real world?[Click]We hit upon an idea that the traffic accidents are happening incidentally.9
10Rear-end Collision Modeling Rear-end collisions are happening by a combination ofVarious brake trigger timingsBrake trigger timing* Various response delaysBrakingBrakingpatterns* Various braking patternsCoastingThis slide that better explains our idea.Rear-end collisions are happening by a combination of[Click]Various trigger timings,which are the timings that drivers notice a collision danger,Various response delays,which are dependent on driver characteristics,And various braking patterns,which are also dependent on driver characteristics.Then,how was the hypothesis verified?How was the hypothesis verified1010
11EDR Data Analysis Source : NHTSA EDR database Speed reduction (mph) 30 CDR system connectedto a crashed vehicleSpeed reduction (mph)(Total 70 cases)30283m/s24m/s5m/s20- 2020Number of cases11- 4010It was verified with actual rear-end collisions.The actual rear-end collision data.were obtained by the analysis of the NHTSA EDR database.The graph on the left hand sideshows a histogram by braking onset.You can see that nearly 70% started brakingless than about 1 second before the crash.The graph on the right hand sideshows the speed reduction by braking onset.As you can see, the average deceleration was3.7 meter per square second.641- 60None-1s-2s-3s-4s-5sNone-1s-2s-3s-4s-5sBraking onset before a crashBraking onset before a crashSource : NHTSA EDR databaseftp://ftp.nhtsa.dot.gov/NASS/EDR_Reports/
12Comparison between EDR and simulation resultEDRSimulation result shows similarity to EDR[%]Simulation result50-30-20-10-40-504030Speed reduction [mph]Frequency2010Assuming that the EDR is equipped with the vehiclesin rear-end collision simulation,the red bars and red broken line were obtained.The comparison indicatesthat the simulation result showsclose similarity to EDR property.Thus, rear-end collision simulation was created and verified.None-1s-2s-3s-4s-5sNone-1s-2s-3s-4s-5sBraking onset before a crashBraking onset before a crash1212
13Rear-end Collision Simulation Without a System+FCW,DBS & CIBWith a SystemWarningQSPRTDBSCIBBrakingpatternsADangerisnoticedlaterthanwarningonsetBrakingwill besoonerCoastingBrakingBCDDangerisnoticedprior towarningonsetEThis sketch shows the rear-end collision simulation without a system.By adding FCW, DBS and CIB effects onto all cases,[Click]A rear-end collision simulation “with a system” was generated.For the cases like A and B,the danger is noticed later than the warning onset,the warning is expected to help the driver notice danger soonerand apply braking sooner.Furthermore, DBS and CIB effects are overlaid onto driver response.For the cases like D and E,the danger is noticed prior to the warning onset,therefore the warning itself is not effective,DBS and CIB effects are overlaid onto driver response.By comparing P with A, Q with B, C with R and so on and summing them up,the system benefit can be estimated.System benefit can be estimated by comparingP v.s. A, Q v.s. B, R v.s. C and so on.1313
14Simulation 1 FCW onset DBS onset CIB onset = 1.7 sec = 0.8 sec = 0.45 secSV = 20m/s*Under the Japanese MLITguideline in 2003Not expectedto pushbrake pedal*Expected topush brake pedal*e.g., sleepy or fell asleepFCW+DBS+CIBCIBSpeedreductionCrashavoidanceratio3.2mph4.5mph56.3mph8.1mph10(mph)15FCW+DBS+CIBCIBThis and the next slides showwhat we got as a result of our ACAT program.The control onsets applied in the simulationis shown in the box on the top.For those who are expected to push the brake pedal,it is estimated that a system with FCW, DBS and CIBcould reduce 8.1mph of speed reduction and 1.0% of crash avoidance.And for those who are not expected to push the brake pedal,due to sleepiness or falling asleep,only CIB is activated and it could reduce 4.5mphbut could not reduce crashes.You may feel that the results show very poor crash avoidance ratios.This is due to the settings shown on the top.These settings are based on the Japanese MLIT guideline in 2003;Ministry of Land industry and transport guideline at that time.As I stated at the beginning of the presentation,the system has been developing year by year.The current guideline permits stopping before a crash.0.0%0.0%0.0%1.0%510(%)1514
15Simulation 2 FCW onset DBS onset CIB onset = 2.0 sec = 1.1 sec = 0.75 secSV = 20m/sNot expectedto pushbrake pedalExpected topush brake pedalFCW+DBS+CIBCIBSpeedreductionCrashavoidanceratio5.7 mph59.1 mph1012.3 mph15.4 mph(mph)15FCW+DBS+CIBCIBThis slide shows the result of another parameter setting.All control onsets were set to engage sooner compared to the previous slide.It is shown that FCW plus DBS contributes crash avoidanceat the speed of 20m per second.Thus, the method makes it possibleto calculate the benefitsfor each function and for each control setting.0.0%0.0 %7.9 %51013.5 %(%)1515
16Conclusions An estimation of the benefit of FCW, CIB and DBS was developed.Driver response before collisions was measured in the driving simulator. It was shown that response varied widely from person to person.Rear-end collisions were reconstructed. The result was validated with the EDR data analysis.Estimation method demonstrated the benefit difference by each specification.This slide summarizes the conclusions.16
17Current Study; LDW benefit estimation Here I show you our current study.Toyota is developing the LDW benefit estimationusing traffic accident data in Japan.For the United States,[Click]Toyota is not familiar with the detailed traffic accident data.We found that Dr. Gabler of Virginia Techis the expert of this kind of research,We are proceeding with the research together.Both works are in the first stage.The results of the studies will be publishedat the upcoming SAE World congress.
18End This is the end of my presentation. Thank you for your attention. 1818