1
`
`Mercedes-Benz USA, LLC, Petitioner - Ex. 1008
`
`

`
`Collision Avoidance Technologies
`
`94C038
`
`Mltsuo Kawai
`Toyota Motor Co.
`
`ABSTRACT
`
`Studies have shown that more than 90 % of all traffic
`
`accidents are caused by driver error involving the recognition of
`the surrounding environment, errors in judgment, or poor
`driving habits. In this regard. automobile manufacturers have
`been developing technologies to increase the ficld of vision.
`to enhance the basic maneuverability of vehicles to ensure
`safe driving. and marketing such vehicles accordngly.
`Moreover, efforts to develop technologies to support the
`driver's recognition. judgement. and operation in a more
`positive manner are becoming increasingly active. in line
`with the recent intensified needs for safety. and advances in
`computation / information processing systems that have been
`developed over the past years. This paper provides an
`overview of the status quo of vehicle collision avoidance
`technologies. and discusses the technological and social
`issues for the practical applications of such technologies.
`
`1. INTRODUCTION
`
`Among the social issues considered recently in the
`automobile realm - such as air pollution, energy resources.
`fuel consumption, traffic congestion. recycle needs, etc. - the
`reduction ofautomobile accidents and enhancement of safety
`are the key issues that automobile users are most interested
`in. In order to cope with such issues, automobile manufacturers
`and governmental agencies (in concerned countries ) have
`been implementing various measures to reduce traffic
`accidents, these measures have shown considerable positive
`effects.
`
`Fig. 1 shows a trend involving the number of fatalities in
`traffic accidents that occurred in several countries, assuming
`that the number in 1970 is l00.[l]
`In each country , the
`number of fatalities was reduced, starting from the first half
`of the 1970s and extending throughout thel980s.
`In Germany,especially. the number of fatalities was
`further reduced front the first half of the 1980s. while in Japan
`
`71 .6(5,373)
`70.0(1 1 .476)
`67.0(1 4,595)
`
`Germany
`41 .7(7,995)
`'00
`
`'00
`
`‘O5
`
`l_-‘ig.l Number of Fatalities in Traffic Accidents
`
`fatalities increased during the 1980s. A similar tendency is
`found looking at the number of fatalities per 100 million
`kilometers of automobile use, the number of fatalities was
`reduced drastically in the 1970s, and it has stayed nearly
`constant for the past several years. From the dramatic
`increase in number of automobiles, it is concluded that the
`development of new technologies for automobile safety. and
`the construction of a new traffic infrastructure are necessary
`for a major reduction in the number of future fatalities.
`in the past, automobile safety technologies were
`developed with the primary focus on passive safety. A
`reduction in the number of casualties and fatalities can be
`expected from improvements in vehicle crashworthiness
`,safety belts, and airbag systems. Therefore, automobile
`manufacttuersaredevelopingsuchtechnologiesand governmental
`agencies are actively encouraging their use, which are
`showing some positive effects
`On the other hand.various types of developments have
`been made in the area of active safety to reduce the number
`of traffic accidents. Namely, technology was developed to
`ensure an adequate field of vision for driving,_ to increase the
`visibility of meters,and to simplify various switch operations.
`
`2
`
`

`
`Improvement of vehicle dynamic performance is also a key
`issue in the area of active safety. Along with improvements
`made in the basic performance of brakes and suspension
`systems. new functions have also been added by means of
`electronics, such as Antilock Brake System (ABS). Traction
`Control System (TCS), etc. Although it is difficult to
`determine the correlation between such active safety
`technologies and the reduction of accidents , they have
`proven to be an effective means of reducing driver
`workload, and preventing driving errors and the induction of
`such errors.
`In addition, steps have also been taken to -
`improve the traffic infrastructure. The addition of traffic
`signals and their systematic control, street lights with higher
`intensity illumination, traffic information display panels,
`etc. have contributed markedly to the reduction of traffic‘
`accidents.
`
`In this paper, present and future collision avoidance
`technologies are studied from the point of view of the driver's
`recognition, judgements, and operations and issues that deal
`with practical applications are discussed, including social
`problems. It is generally stated that more than 90 % of traffic
`accidents are caused by human errors. Fig. 2 shows rates of
`various types of human errors that cause accidents. [2] This
`figure illustrates that'errors in recognition and judgment are
`major causes of traffic accidents. In the collision avoidance
`system, exterior environment sensors and information from
`dteinfiasutntmeaemilizedtosripplernetithmutn recognition. and
`judgment, and to automatically compensate for human errors
`
`automation directly to automobiles . The automobile driving
`environment is quite complex. which necessitates highly
`capableenvironmerlalsensorstopropetlycollect information, As
`long as human beings want to drive automobiles attheir-own
`discretion, significant fluctuations are apt to occur in their-
`driving functions.
`In order to attain a proper system of
`cooperation between human-beings and machines undersmh
`circurrstances, we must not only overcome technological issues
`involved in the human machine interface (I-IMI), based on
`human factors,but also strive for the creation of social
`consensus and resolve the issues related to product liability,
`etc.
`
`2. STATUS QUO& DEVELOPMENT TRENDS
`OF COLLISION AVOIDANCE TECHNOLOGIES
`
`Table I shows the present status and development trends
`of technologies relating to preventive safety such as
`collision avoidance.
`Items related to safety include the
`acquisition of information from outside the vehicle, the
`methods to influence driver performance, and the means of
`monitoring driver and vehicle conditions. Items related to
`the monitoring of vehicle conditions will not bediscussed in
`this paper.
`
`DEFINITE CAUIAL
`FACTORS
`
`c/mam. an
`BEVERITY-DIOHEAIING
`E DEPINITE on taneouue
`FACTORS
`
`ON at-re : Ii-SIBI nucleon-In
`IN DEPTH 1 NDARO Acfildlflll
`
`ON
`IN
`SITE
`DEFTH
`RECOGNITION
`ERRORS
`
`ON
`IN
`RUE
`ISEFTIJ
`IIECISION
`EHIHCRS
`
`ON
`IN
`SITE
`DEIVTD-I
`PERFORMANCE
`canons
`
`ON
`IN
`CIT‘
`DEPTH
`91-Haas
`
`Fig.2 Percentage of Accidents in which Human Errors were identified as Causal Factors.
`
`Such automation technologies used to eliminate human
`errors have brought about dramatic effects in various sectors
`of manufacturing. Semiconductor wire bonding and the
`mounting of electronic devices to printed circuit boards are
`good examples of the above. As will be discussed in this
`paper, however. it will be difficult to apply the scenario of
`
`Technologies to assist recognition, judgments, and
`operations of the driver , as well as those to monitor his or
`her condition, are focused mainly on the relationships amont
`the driver. the vehicle and the vehicle driving environment.
`The present status of safety related technologies will be
`discussed. using several preventive safety systems shown
`inTable l as examples of such technologies.
`
`3
`
`

`
`Tablel. Status and Development Trends of Technologies related to Preventive Safety
`
` Already tnerohandieod existing product: Product: to be moruhondleod. in near lulure
`
`3- Forward Monitor
`‘
`
`- High Luminance Head Lamp
`small Distortions Windehield
`
`- Night Vision
`Light control
`
`- Ultra Violet Lamp
`
`2- Sldewerd Monitor
`3- Hear Monitor
`
`- clearance sonar
`- Rear View Monitor
`
`- Back Sonar
`
`3. “mo.-m.m;..
`'
`in opnmuon.
`Blake oi
`: cg||i.|°n
`
`- Laser Radar Collision Warning ‘
`- SH; Radar Oolllelon Warning
`.
`- SI-IF Radar Slind Spot Monitor
`, sup.’ High Fruwnwy
`
`.
`
`1- Supplementation
`
`- Antilock Brake System
`- Traction control System
`
`- Lune Mark Deviation Warning System
`
`- Lana Change Warning system
`
`5- Automation
`
`' Automatic Transmission
`
`' lnteligent Ct-use Control
`
`- Auto Brake
`
`- Drunken Driving
`- Drug
`- Droweinoee
`[F.|igu.)
`
`- Breath Alehohctl Ignition
`Interlock Devices
`
`- solely Drive Adviser
`
`- Preventing Drowsiness uelng
`Image Processing to detect Blink
`
`
`
`Monitor.WuningandAssist
`
`
`
`Monitorcondition
`
`2.1 Systems to Assist Recognition
`
`Allowing the driver to accurately recognize his environment
`constitutes the basis of preventive safety. Basic technologies
`to provide accurate recognition ensure an adequate field of
`vision in the vehicle. That is, they are technologies to ensure
`appropriate locations of A, B and C pillars and a proper
`driving position (the position of eye point), and to optimize
`lights and the windshield wiping area. Ensuring the rear
`field of vision when the vehicle is backing up is one issue that
`may cause potential complaints, such as difficulty and
`uncertainty in operation. though severe accidents such as
`fatalities are not likely to occur because of the low speeds.
`Fig.3 shows traffic accidents which occurred at night.
`[3]The number of fatalities occurring at night is about 30 %
`greater than the number in daytime. In general, the rate of
`accidents in rainy weather is about 3 to 5 times higher than
`that in good weather. They are presumably due to the
`lowered activity levels ofthe body and mind at night. and the
`deaeme ofrecognition ability in the dark or rainy environment.
`Technologies to allow the observation ofobjects in such
`difficult visual conditions have been developed.
`o
`.
`l)1l«Ei:2.n4’i‘:l11‘<3>‘vTs%(e)Tp:)<:ifications of Toyota's Rear View
`
`Monitor. The compact CCD camera installed at the top of
`trunk lid is used to show the rear view on the display when
`the vehicle backs up. This systems shows objects in the
`vehicle's blind spot, which is highly effective. Improvements
`are needed, however, in the following areas. 1. Distortion of
`images due to the wider lens angle , 2. visibility of objects at
`night. and 3. cost.
`
`' V
`.
`
`.
`-'
`
`--
`
`x\
`‘ \
`,
`_
`,
`,
`‘ V ‘
`
`,
`
`,
`
`1 —1
`K - S
`‘
`1:
`
`I
`""1""
`
`nlelleunelehlel ldldthlu.
`byulqen-L
`
`j
`
`-
`
`"
`
`‘
`
`'
`
`Mun-clc'IIItaly
`
`Fig.4 Rear View Monitor
`
`4
`
`

`
`The Back Sonar and Clearance Sonar are aimed at
`
`detecting the distance to each obstacle by means of ultrasonic
`sensors installed at the rear and comers of the vehicle, and
`informing the driver of the distance to the obstacle by
`changing the tone of the alarm according to the distance. Such
`devices are equipped on many vehicles{'IOYOTA:PREVIA) in
`Japan, which are effective devices to assist drivers during
`parking maneuvers.
`
`2) Light Control
`Figs. 5 show the light control actuator developed by
`Koito. This system senses the presence of on-coming
`vehicles in the opposite lane, or vehicles operating ahead of
`the
`by means of radars and infrastructure information,
`and controlstheshadeposition properly so that thc cutlinc (upper
`boundary of illumination) against such vehicles can be
`minimized to avoid glare. When the cutlinc control is done
`to a light of a fixed cutlinc. the distance to recognize
`obstacles within the vehicle's operating lane is dou blcd when
`vehicles are coming from the opposite direction. which makes it
`possible to increase the safety margin of drivers in making
`proper judgments and operations to avoid accidents.
`
`cut|ine(upper boundry ol
`illumination)
`
`Vertical
`
`(100W)
`
`Horizontal
`
`controled separately
`right and tell
`
`Fig.5 Light Control Actuator (Koito)
`
`Volvo developed a headlamp which utilizes ultraviolet
`light. [4] It is capable of recognizing objects from a distance
`of 100 to 150 m, compared with a distance of 40 to 50 m in
`ordinary passing conditions, if such objects are fluorescent.
`Pedestrians clothes (including jeans) made of cotton or
`nylon normally contain fluorescence. Thercl‘ore, it can
`effectively recognize pedestrians. Since ultraviolet rays
`have better penetration characteristics than visible light in
`severe environmental conditions such as rain, fog, snow, etc.,
`the ultraviolet light produces excellent results in recognizing
`pedestrians, obstacles, lane markers, etc. which contributes
`to traffic safety. General Motors (Delco Elec.) proposed the
`use of an infrared camera , called Night Vision, as a system to
`supplement the normal field of vision at night.[5]
`The effectiveness of systems to assist in recognition can
`be understood easily by drivers since they are extensions of
`me_astn'es to improve the field of view. For some of them ,
`
`it is difficult to determine the direct effectiveness
`however,
`in reducing accidents, and costs of on-vehicle came;-as‘
`displays, light control actuators, etc. , are prohibitive. 170,
`these reasons, such systems have not been used extensively,
`When they are put into practical use, however, drivers of
`vehicles equipped with such systems would be able to
`recognize surrounding conditions, even in severe Conditions
`such as foggy weather, and they may drive their vehicles at
`higher speeds than in a normal situation, which may cause 3
`new safety issue. Therefore, developing products need to takg
`into consideration not only technological issues , but also the
`potential for adverse effects on the entire traffic environment.
`
`2.2 Systems to Assist Judgements
`
`Various types of systems to detect and judge potential
`hazards caused by errors of drivers in recognition or judge-
`ment, and to warn such drivers by means of alarm devices,
`etc., have been proposed to induce drivers to drive safely.
`
`1) Rear-end Collision Warning Systems
`Fig. 6 shows the rates for different types of accidents for
`all of Japan. [6] For all of Japan, straight crossing path
`collisions account for the highest rate of about 28 % ,
`followed by rear-end collisions, which account for about 24
`% of the total. Fig. 7 shows the characteristic features ofrear-
`end collision accidents which occurred on the highways. [7]
`In the case of highway accidents, however, rear-end
`collisions account
`for the highest rate of 37%. Since
`approximately 70 % of the rear-end collisions on the
`highways occurred on straight sections of the road, the
`drivers could have failed in recognizing of the state of the
`vehicle ahead. A proper measure to cope with this problem is
`not only to make invisible objects visible. but to alert the
`drivers to the presence of the objects, which is an area
`machines can handle relatively easily.
`
`Collision with llxed oljogtrtg nonculllslcn accidents
`
`collision vrlth pedestrian
`
`Straight crossing path
`27.9%
`
`collision with Item
`vehicle In Transport
`
`“°""2‘5‘,1°"“i ‘
`
`Roar-anti Collision
`24.1%
`
`Turn across
`13.1% p‘
`
`Fig.6 Rates of Different Forms of Traffic Accidents
`
`5
`
`

`
`collision with axed object or
`monoolllelon oocldonla
`
`collision with pedoatrlon
`or pedoloyclo
`
`Rear-end OOIIISIOHT We
`
`Near collision
`17%
`
`W
`
`Collision with Motor Vehicle in Traneport
`
`Fig.7 Characteristic Features of Traffic Accidents Occurring on the Highways
`
`Systems using laser radars are marketed in Japan as rear-end
`collision avoidance systems. as shown in Table 2.[8][9][l0]
`Infrared laser radar is capable of detecting objects over a
`distance of 100 m or so against a reflector installed on the
`rear of a vehicle. The measuring accuracy is i 2 - 4 m,
`which is sufficient for the vehicle-to-vehicle distance of 40-
`50 meters.
`
`{.
`
`to-vehicle distance‘, and "driving at lower speeds than
`usual". The effects of the system on the drivers, other than
`that on their driving habits, were as follows. "Feeling more
`secured", "mentally feeling easier", "drowsiness is gone",
`"the warning sound is like a partner to accompany me“, etc.
`according to statements made by the drivers.
`
`Table.2 Rear-end Collision Warning Systems using Laser Radars(marketed in Japan)
`"”“s“‘°‘S“‘ ““°"°'
`for passenger car 8-large truck)
`D stance.
`arning
`(Oot.'92)
`
`for far e truck)
`,Tra lc Eye
`(Dec.'89)
`'
`
`for lar o truck
`Safety Eye
`May.'92
`
`"""5°' Rad“
`
`' W1 94><H87>< D1 20
`mm
`
`- W200><H75><D'116
`l'T'li'l'l
`
`- VV200><H52>< D1 40
`lT'll"l"|
`
`peak power
`I'ange/
`
`- 15w
`
`2. System
`
`"""“"'" ""9
`concept
`
`Suppression of erroneous
`operations.
`The distance to recognize
`hazards ls ehorbwith many
`restrictive conditions
`against warning .
`
`- 14w
`
`-25w
`
`- Max : 100m/d:-4m
`
`- Max : 99m/:l:2rn
`
`Warning with a proper
`balance for the suppression
`of too-frequent sounding or
`erroneous warnings under
`the current driving
`conditions.
`Three levels of
`“l‘ar"."rnedIum" and "near".
`
`Warning get on the eater
`side.
`
`Sounds too trequently and
`Irritating.
`
`Three levels of
`"far".“m edlum“ and "near".
`
`Fig. 8 shows results of evaluations done by the users of a
`warning system developed by Nissan Diesel. [ll] Vehicles
`equipped with this system were large and medium size
`trucks. One hundred seventy one drivers were employed for
`the evaluation . Drivers in their40s or over accounted for 81
`% of the total. and a majority of the drivers had more than 20
`years of driving experience. Seventy five percent of the
`- driver's trips were 600 km or longer, with frequent useof the
`highways. About 66% of the drivers used the rear-end
`collision warning system for one year or longer. More than
`82 % of the drivers stated that their driving habits changed
`after the system had been installed - namely, "they became
`more careful while driving", "trying to keep a longer vehicle-
`
`As for the effectiveness in prevention of rear-end collisions,
`some drivers reported cases in which accidents were
`prevented when they become aware of potential hazards via
`the warning signal and accordingly made the proper steering
`or braking maneuvers. The quantitative evaluation was,
`however, inadequate, as the number of vehicles was low and
`the length of time spent for the evaluation was short.
`Problems related with laser radars are :1) the poor
`performance in
`severe environmental conditions (the
`lowered detection capability caused by rain, fog, snow, etc),
`and 2) relatively long computation time and low accuracy
`caused by differentiation of distance data. and so on. There
`are safety regulations for laser based products to ensure the
`
`6
`
`

`
`0
`
`25
`
`Driving rnorocaroiul
`
`_
`K.
`w
`
`T—
`jj
`
`.'l..'“."".t..'.';...'°""""""
`
`Driving wih confidence
`It has become
`unnucaa to pnytoo
`llllch lite on to
`surrounding eondlllont ‘
`
`- Fig.8 Change of their Driving Manners
`after the Warning system had been installed
`
`safety of eyes. It is necessary to meet the requirements of
`Class 1 (as specified by the regulations) in order to use them
`on ordinary passenger cars, since the users do not have
`professional knowledge of lasers. In order to attain Class 1
`certification, optimization of parameters such as pulse
`frequency and "pulse width. in addition to the luminous peak
`power were necessary. The regulations were revised in 1991
`(the allowable output was increased by threefold), allowing
`the detection capability to be expanded.
`In the USA, rear-end collision warning systems using 24
`Gllz microwave radars have been marketed. They include
`the short distance blind spot monitoring radar system for
`school buses developed by Delco Electronics in 1992, and
`the rear-end collision warning system for buses developed by
`VORAD in 1993.
`~
`Toyota is also considering a collision warning system
`using millimeter-wave radar. lt is a 60 GI-lz FM-CW system
`equipped with a steer mechanism to improve the detection
`performance on curved roads. The warning logic is basically
`the same as that of the laser radar collision warning systems
`now on the market. and the warning timing is calculated by
`the vehicle speed, vehicle-to-vehicle distance. and the
`closing speed. Fuzzy logic was also studied so that the
`warning tinting becomes closer to the timing actually
`perceived by the driver. Problems with millimeter-wave
`radars are the cost and a difficulty in installation (due to
`antenna size).
`In Japan. regulations are such that it is not
`allowed. at present, to generate an adequate output to detect a
`vehiclentruting 30-60 meters ahead. Nevathelem. assignment of
`millimeter—wave bandwidth to automobile applications is
`being considered at the moment, and the practical use may
`become possible in the future.
`'
`
`2) Lane Change Warning System
`Fig. 9 shows a prototype of Toyota's rear/side monitor
`(millimeter-wave radar) system. The radar is capable of
`monitoring objects located up to 15 m behind the vehicle at
`60 GHz, and warning of the presence of vehicles approaching
`
`from the rear, on right or left sides, or during lane change
`maneuvers. There are 3 warning modes. Wamingsare given
`for the conditions shown on the map indicating the vehicle.
`to-vehicle distance . R. and the approaching speed, V;-,
`according to the proximity of the approaching vehicle, as
`shown in the figure.
`
`Warning Area
`
`(A) overtaking vehicle with a high speed
`
`(8) Over taking vehicle upon lane change
`( Sig nul on I Large steering angle)
`
`0 Warning
`
`(C1 Adjacent vehicle In blind spot
`
`Fig.9 Rear/Side Monitor system using
`Mi llimeter-wave Radar
`
`Fig 10 shows the tearlside vehicle recognition technology by
`means of image processing. This system is ‘capable of
`detecting vehicles operating in the adjacent lanes which are
`immediatclyldtorriglttofdtevetficleartdthose approaching from
`behind. by means ofimage processing by two CCD cameras
`embedded in the side mirrors.
`It is capable of such a
`detection even at night. as long as those vehicles have their
`lights on. The detection cannot be done in some cases where
`adequate contrast cannot be provided due to back-light or
`heavy rain, etc. In such a case. the detection incapability is
`judged automatically and shown on the display panel.
`
`’ Fig.lO Rear / Side Monitor System Using Image
`Processing
`
`3) Lane Keeping Warning System
`Table 3 shows a method of an automated lane marker
`
`detection by means of image processing. [l2][l31[14]
`
`7
`
`

`
`Simple differentiation, space filter, and correlation are used
`as methods to detect the lane edge marker. but the space filter
`method and correlation method are better than the simple
`differentiation method in terms of performance. Although
`[me markers may be detected consistently by this method, to
`trace them from a short distance, with less fluctuation of
`images on the display. to a relatively long distance. and the
`erroneous timing calm by the lead vehicle's close proximity to
`the lane edge maker . etc. tends to be problem on this method.
`The detectable distance is 40 in or so in daytime, but ensuring
`good performance in rain or at night remains a difficult issue.
`
`completely with the desire of the drivers. At present,
`however, the situations are quite insufficient. A huge cost
`will be involved for the installation of an adequate number of
`sensors, which makes it important to form a proper social
`COHSODSUS.
`
`Another problem of warning systems is’ that proper
`accident avoidance operations by the drivers are indispensable
`after a warning is given.
`In other words, the expected
`effectiveness of such a system may not be achieved if" the
`driver's response time and the subsequent action is notpnoper.
`In consequence. some assist system (for automatic operations)
`
`'l‘ablc.3.Currcnt Status of Lane Keeping Technologies
`
`Type
`/ method
`
`Template
`matching
`
`N ‘SSA N E go detection
`pvs
`(P9|’3°"3‘ V9'''5¢' ; FUJlTSU
`+ edge tracing
`S stem
`ADA
`(Acuv, guys
`Asslst Bynem)
`MOVER-3
`(Mann 9 V. ‘Ion
`Exporlmental
`Flobot-3
`
`3-dimensional lane
`3 U BA H U geometry recognition
`by flereograph
`Binary procosstn «-
`aoalmmauon 0‘ ma
`MAZDA among lndlvldual lines +
`method least : uares
`
`Processing
`Outline of
`P°"°'ma"°e SW96
`processing
`capable of detection
`Templates for edges
`ever with unclear lane
`TOYOTA + templates lor
`two edge mam marks found at present
`ggtoagapejgfloxlmation by
`Egg. dgggoflgn
`lnsrfensitiI\/efitio road
`+ "°:g::,,,,,.,, T°*°TA . a::.;:. ...$.......
`32.23:." "a"°°
`Capable or detecting
`lane mark at short
`distances
`: ;- rox.25rn max
`Tr" m°m°d °' lea"
`Squaw‘ '3 gimp“ ‘°
`affected by noise
`caloulate.but tends to be
`surface luminance
`insensitive to road
`varlauon
`
`‘°°"~=°°
`
`'
`
`1 Oomsec
`
`30msec
`
`Issues to
`deal with
`. P'°V°m'°” °' V°’“°‘°
`saga erroneous
`detection
`To manage speed
`|_’:'_|°a‘r3|“’9"m°" 0' '3' '3'“
`sm»--==auo» « em
`radius detection
`Ftaoognltlon ot tar lane
`marks
`slze reduction of
`equipment
`Stabilization of linear
`approximation
`To mamas‘ speed
`To cope with road
`variation
`wd°°° '"m'"°"°°
`
`‘
`
`Some of the systems to assist judgments in general. and
`the rear-end collision warning systems in particular. have
`already been put into practical use with considerable
`effectiveness. They are not adequate, however. in terms of
`reducing all types of automobile accidents, as they are set for
`a limited number of users (commercial truck drivers).
`If
`they are to be used extensively on passenger cars as well, it
`will be necessary to have such systems matched with havard
`judgments from a number of drivers. Some drivers are
`irritated by repeated warnings in normal driving conditions,
`if such a system is set on the safe side. If it is set at the limit
`value to suppress the activation of warning, on the other
`hand, the effwtiveness in hazard avoidance may drop clue to
`the delayed response of the driver. In such regards, the most
`crucial issue we should deal with in the near future will be to
`
`improve the quality of warnings (by eliminating no-warning
`failure and reducing erroneous warnings). it must be generally
`noted. however, that each driver must make judgments
`based on not only the information from the lead vehicle, but
`also those of several other vehicles operating ahead, while
`paying attention to the presence/absence of vehicles
`operating behind and the traffic environment (on signals,
`branching/merging roads/lanes, etc.). Therefore, a number
`of sensors are necessary, not only on the vehicle. but also in
`the infrastructure, in order to have such a system match
`
`linked with this type of system is presumably necessary for
`furtlter reduction of accidents.
`
`2.3 Systems to Assist Operations
`
`In this area, there are systems to assist the driver's vehicle
`operations and those to automate the vehicle operations. The
`former systems are aimed at limiting, to the safe side. the
`operations done around the vehicle's maneuvering limits ~
`ABS (Antilock Brake System), TCS (Traction Control
`System), etc. are already used widely. It is indispensable for
`those control systems to detect driving operations near the
`vehicle maneuvering limits, and to accurately inform the
`driver of hazardous conditions.
`
`The latter systems, on the other hand, are aimed at
`reducing the driver's loads in recognition and judgment by
`partially controlling the driving operation, and preventing
`accidents that may be caused by erroneous operations. These
`systems have not been adequately marketed as in the case of
`the former. The development status will be described in the
`following. using ICC (Intelligent Cruise Control) and auto
`brake system as examples.
`
`1) ICC
`ICC is a system in which a headway distance control
`
`8
`
`

`
`where theactivation of the sensors would be inappropriate (it
`was considered that the activation of ICC would be hazardous
`under poor visual conditions in which the driver could not
`see properly)", and "the system would operate only on
`highways and straight rural roads". According to results of
`an informal test, it was proposed that "the headway time
`should be adjusted by the driver" and that "it would be
`necessary to standardize the ICC operation procedure to
`prevent driver confusion".
`
`2) Auto Brake System
`Auto brake systems are aimed at predicting possible
`collisions, and the automatic control of braking. It may be
`said that they are systems to avoid accidents more directly
`than the ICC. The primary objective of Mazda's Auto Brake
`System is to compensate for the driver's errors, and was
`developed based on the following concepts.
`(1) It should not interfere with normal driving operations,
`(2) It should induce the driver to make hazard avoidance
`action himself/herself by giving a warning if the dangerofa
`collision is predicted.
`(3) Emergency braking would be activated only when a
`proper harard avoidance action per (2) above was not taken by
`the driver.
`
`(4) It would be used on highways (vehicle speed: 100 km
`maximum, curvature R: 300 in minimum) only.
`The system configuration is shown in Fig. 12. The
`headway distance sensor is a scan-type laser radar (view
`angle = 23dcg x 20deg. range =120 in), using a yaw rate
`sensor and a steering angle sensor.
`
`Fig} 2 Mazda Auto-Brake System
`
`The current status of ICC and auto brake system
`developments have been described in the foregoing.
`It is
`possible for both of them to noticeably reduce errors in
`recognitions, judgments, and vehicle operations compared to
`human beings, owing to their ability to assist driving
`operations by automatic control.
`in this regard, more
`positive effects on the avoidance of collisions can be
`
`function is added to the conventional cruise control. This
`added function controls the acceleration of the vehicle based
`on the headway distance and the relative speeds of the lead
`vehicle, in order to prevent a collision accident.
`
`driver commands
`
`driver '
`
`(cruise cdntrol
`buttons)
`
`calculate
`‘saw.
`speed
`
`range rate
`
`distance
`midway
`sensor
`
`‘em’:
`59”
`V driver
`—
`
`d:i|t?ecdi:ne°e'd
`.
`9.
`tor cruise control
`
`V—°°
`CIUISG
`control
`
`Fig.1] Leica Intelligent Cruise
`Control - Overview
`
`If
`Fig. 11 shows the ICC system for Lcica. [l5][l6]
`there is no load vehicle (target), the control uses the vehicle
`speed set by the driver as the desired vehicle speed, as in the
`case of conventional cruise control. If there is a lead vehicle,
`an IR headway distance sensor (laser radar. view angle =
`Sdeg x 3deg. range = 150 m) is used to detect the distance to
`the lead vehicle and to change the vehicle speed so that the
`distance remains constant. The actuator for the vehicle speed
`comrols the engine (throttle) and transmission (shift down)
`only and does not control the brakes, similar to conventional
`cruise control.
`
`Table.4 Data for L/cica ICC
`
`- ll‘-l time ol llight measurement
`- Monobenm
`[E513]
`
`§m
`
`I1—
`§IE_
`W
`
`arget acqutstt on
`
`river set speed
`- Onloll
`- Tar el lracltln
`- automatic
`- only vehicles with a speed
`at > 30% at the ice at velocity are
`attired as a mat
`
`2-
`
`III
`
`P0
`
`)
`>-
`in
`
`Table 4 shows the specifications of the system. For
`determining the specifications, the following preconditions
`were set; "the system should not activate under conditions
`
`9
`
`

`
`expected from those systems than other assist systems. On
`the other hand, as high performance and reliability are
`expected of such systems. operating conditions of the
`systems or roadway conditions are often restricted, such as
`the above-mentioned systems ( many ICC or auto-brake
`systems operate only in good environment or on highways ).
`To improve the performance of these systems, some
`headway distance sensors ( millimeter wave radar [18].
`which can operate in all weather conditions, etc. ) , lane
`marker sensor [l9]( to detect the target lanes using computer
`vision ),and ICC systems with brake-control [20] were
`developed. To provide better performance and reliability,
`the use of the infrastructure is expected. [20] But, for all
`these developments, there are many issues that should be
`dealt with such as countermeasures to be taken when a failure
`occurs. product liability, etc. .
`Moreover, a number of issues must be solved in terms of
`the human-machine interface. In addition to the issues, such
`as the ICC operation procedure, etc.
`,described in the
`foregoing. possible delays of the driver's hazard avoidance
`actions are also a cridcal issue that has to be dealt with.
`It
`may be presumed that such delays are caused by a lower
`concentration on driving, since the driving becomes relatively
`monotonous due to automation, or by the overconlidencc in the
`automated systems. Therefore. cooperative designs taking
`account of such delays becomes necessary.
`
`2.4 Driver Monitoring Systems
`
`The systems described so far are those related with
`vehicles and the driving environment. but the relationship
`between the driver and vehicle is also an important item for
`collision avoidance. Typical systems in this area are driver
`monitoring systems. Such systems detect. reduced driver
`capabilities, and warns the driver or restricts the vehicle
`motion in one way or another to avoid accidents. Current
`status of development efforts, including the detection of a
`driver's drowsiness and the detection of breath alcohol, will
`be described in the following.
`Table 5 shows various methods to detect the driver's
`
`drowsiness. [21]. As a whole. the method to use steering
`angle signals accounts for the majority of the work [22][23],
`but the method to analyze the vehicle behavior in the lateral
`direction of the lane by means of computer vision, and to
`detect it as "vehicle's meandering‘
`is also available [24].
`The method to