SCHOOL OF MECHANICAL ENGINEERING

DEPARTMENT OF AUTOMOBILE ENGINEERING

SAU1603 – AUTOMOTIVE SAFETY

UNIT IV – COLLISION WARNING AND AVOIDANCE

 UNIT IV
COLLISION WARNING AND AVOIDANCE

 COLLISION WARNING SYSTEM

 CAUSES OF REAR END COLLISION
 FRONTAL OBJECT DETECTION
 REAR VEHICLE OBJECT DETECTION SYSTEM
 OBJECT DETECTION SYSTEM WITH BRAKING SYSTEM INTERACTIONS.

1. COLLISION WARNING SYSTEM

Collision Warning with Auto Brake is an active safety system that helps the driver to
avoid or mitigate rear-end collisions.
It uses forward-looking sensors to detect obstacles ahead of the vehicle.
When a high risk for a rear-end collision is detected the system helps the driver by
providing a warning and brake support.
If the driver does not react in time and a collision is judged to be unavoidable, the
system will automatically brake the vehicle.
This may not avoid the accident, but the consequences can be reduced.
This system is introduced in two steps. The first generation, called Collision Warning
with Brake Support, is currently on the market in the new Volvo S80 allowing activation
on vehicles that are moving or have been detected as moving. The second generation,
called Collision Warning with Auto Brake
Collision Warning with Auto Brake where the area in front of the vehicle is continuously
monitored with the help of a long-range radar and a forward-sensing wide-angle camera
fitted in front of the interior rear-view mirror.
A warning and brake support will be provided for collisions with other vehicles, both
moving and stationary.
If the driver does not intervene in spite of the warning and the possible collision is judged
to be unavoidable; intervention braking is automatically applied to slow down the car.
This aims at reducing impact speeds and thus the risk for consequences.

Figure 4.1. Schematic of collision warning system.

Components of collision warning system:
 Sensor System
Information about the traffic situation in front of the host vehicle is obtained from
two sensors:- A 77-GHz mechanically-scanning forward looking radar, mounted
in the vehicles grille, which measures target information such as range, range rate
and angle in front of the vehicle in a 15 degree field-of-view.
A 640*480 pixel black and white progressive scan CMOS camera, mounted
behind the windscreen, which is used for classifying the objects, e.g. as vehicles,
in a 48-degree field-overview. Since the camera is used for reporting both vision
objects and lane markings, the field of view was chosen to work for both.


Figure 4.2. Collision warning system in vehicles.

 Collision Warning
The Collision Warning (CW) function is targeting to avoid or mitigate collisions
by means of warning the driver ahead of a possible collision.
The system requires high usability, low number of nuisance alarms and an
efficient Human Machine Interface (HMI).
The Collision Warning system should provide a relative late warning in order to
reduce nuisance alarms and to reduce the possible misuse where an early warning
system may build a trust that is falsely interpreted by the driver to allow for
execution of non-driving tasks. The activation of the Collision Warning will
therefore approximately occur when the driving situation is considered to be
unpleasant.
However, it shall allow the driver to brake to avoid or mitigate an accident
 provided the following distance was initially longer than the warning distance.

Functions:
Threat Assessment
 The aim of the threat assessment is to understand if the information
from the forward sensing system shows that there is a risk for collision
 The first step is to approve a lead vehicle as staying in the forward path
within a given time to collision utilizing intra-vehicle and yaw-rate
information.
 Given an approved lead vehicle a second step calculates a total warning
distance, i.e. the predicted distance required for avoiding a collision.
 The total warning distance base calculation is derived from a sum of three
distinct distance calculations.
 The first is the driver reaction distance which is obtained from the
predicted driver reaction time multiplied by vehicle speed.
 The second is the system reaction distance which is obtained from the
system reaction time multiplied by vehicle speed.
 The third is the braking distance to avoid impact using the current physical
states of the lead vehicle and the host vehicle using the constant
acceleration model for the behavior of the host and the target vehicle
closely mimicking the CAMP late warning algorithm.
 The sum of above provides a total warning distance.
 If the distance to the forward vehicle becomes lower than the total warning
distance a warning is to be issued.

Collision Warning HMI

 An efficient HMI(Human Machine Interface) for a warning system is
characterized by a low driver reaction time, as this is crucial for improving
the possibility for the driver to mitigate or even avoid a collision.
 An efficient HMI puts requirements on low false and nuisance alarm rates,
since there is a risk for overexposure that may lead to drivers deactivating
the system.
 The warning interface is a dual modality warning incorporating visual and
audible channels.
  The visual warning is a flashing red horizontal line located in the lower
part of the windshield in the forward direction of the driver.
 The sound consists of tone burst with harmonics content.
 When the audible warning is active the sound system is muted. The
Collision Warning can be turned off by a main switch.
 The system includes a warning distance setting using three levels. The
levels have been defined by balancing driver behavior in late brake
situations versus normal driving behavior.
 The warning distance settings are differentiated by the deceleration level
used in the different settings the predicted brake ability by the driver.
 They also reduce warnings in normal driving situations to different levels.

 Figure 4.3. Collision Warning head-up display



Driver Over ride:
 The objective of the driver override function is to inhibit a brake intervention
when the driver has the situation under control. However, this is difficult or even
impossible to measure and therefore driver inputs as steering and braking
activities are considered instead, as these are the natural countermeasures in a
collision event.
 The release of the accelerator pedal is considered, as this indicates that any further
acceleration is undesired, and it can be assumed that the driver is thereby
acknowledging a collision risk.
 Since the level of action that is required to activate a steering or brake override
depends on the driver and on the traffic situation, the decision threshold is
empirically determined through extensive testing in real life traffic situations with
a large number of drivers.

Auto Brake
 It is beneficial to the driver to get support in the upcoming collision event. This
can be achieved by reducing the collision energy by optimizing driver initiated
braking or through automatically putting on the brakes prior to the collision
event.
 When providing autonomous interventions that override or complement the
driver’s actions, one has to ensure that customer satisfaction is not negatively
affected by false interventions.
 Customer acceptance is crucial in order to increase take rates and thus to increase
the overall real- life safety benefit of the system. It is therefore necessary to
implement a decision making strategy that reduces the amount of false
interventions while not missing collision events where the driver needs support.
Therefore, an intervention decision should be based on two main information
categories: traffic situation data and driver actions. The traffic situation data is
used to quantify the risk for a collision event, in other words a threat assessment
is performed.
 This assessment will never be perfect as sensor information is usually a subset of
the totally available information and mostly affected by latencies. So, a collision
may appear to be unavoidable but is in reality avoidable. Hence, a driver that
takes distinct steering and/or braking action is judged to be in control of the
situation and should be trusted. The driver override function is to detect these
distinct driver actions.
 As soon as the support system has performed the threat assessment and driver
override detection, the outcome can be weighted by the brake intervention
strategy and a decision on an autonomous brake intervention can be taken
Functions of collision warning:
System functionality:
 Alerting the driver
 Braking control
 Restricted steering Driver functionality:
 Changing lanes
 Turning the system on and off
 Approaching another vehicle or Non
vehicle obstacle

2. REAR-IMPACT COLLISION WARNING SYSTEM (RICWS):

1. Common factors that contribute to rear-end collisions include by driver inattention or

distraction, tailgating, panic stops, and reduced traction due to weather or worn
pavement.
2. Rear impact crashes are the most frequent type of bus accidents.
3. Transit buses are particularly susceptible to rear impact collisions because of their
frequent stops, which often occur in traffic lanes.
4. The majorities of bus collisions occur while the bus is decelerating or stopped.
5. The preponderance of crashes occurs with buses stopped during daylight hours, in good
weather conditions, while traversing a straight path, and with the striking vehicle
attempting no avoidance or corrective action.
6. Rear-end collisions are common accident scenarios and a common cause of these accidents is
driver distraction and thus not reacting in time.

3. FRONTAL OBJECT DETECTION

Figure 4.4. System overview of frontal object detection

 Figure 4.5. Schematic of frontal object detection system

SYSTEM OVERVIEW
 The Sensor system has two cameras that can detect vehicles in the medium and far range are installed by
the side of a rear-view mirror and at the ceiling above the back seat and the two sonar sensors that can
measure the distance in the near range are installed at the front and rear bumpers.
 Because the environment of the vehicle changes relatively fast as the speed ofa ego-vehicle is high, we
acquire 2 images of 1 field with 640×240 for avoiding the motion flow and use 320×240 image by sub-
sampling and acquire 2 signals of sonar sensors successively. 

Determination Of The Day And Night Times
 The environment of a moving vehicle greatly varies, it is difficult to detect vehicles
using a single feature or pre-established vehicle templates.
 Various features or templates can be used for various environments, and by
applying the appropriate algorithm to the current environment the day or night time
can be determined.

Vehicle Detection in the Day Time
 The detection system in the day time consists of 4 parts, the preprocessing module
working on the input raw image, the vehicle candidate extraction module by a
shadow region and a template, the validation module by a prior knowledge, and the
fusion module for fusing sonar and image data.
1) Preprocessing:
2) Vehicle Candidate Extraction:
3) Vehicle Candidate Validation
4) Vehicle Detection Using Sonar Sensors
 Figure 4.6. Determination of the light condition

Figure 4.7. Schematic of Vehicle Detection System in the day time

 Figure 4.8. Vehicle Candidates Extraction

Figure [Link] symmetry rate and edge angle map

Figure 4.10. Vehicle detection at near distance

4. REAR VEHICLE OBJECT DETECTION SYSTEM
 Rear object detection systems monitor a specific area behind a commercial motor vehicle,
detect objects, and provide warnings to drivers when they are approaching an object
behind the vehicle while in reverse. These systems assist the driver in avoiding collisions
during backing or parking maneuvers.
 Rear object detection systems detect moving and stationary objects located within a
specific area behind a commercial motor vehicle while it is backing up. Currently
available systems can detect objects within a range of approximately 10 to 20 feet behind
a vehicle.
 They can be integrated with other sensors, such as side object detection sensors to cover
other blind spot areas around a vehicle.
 Audible and/or visual distance-based alerts that vary depending upon the closeness of the
vehicle to an obstacle are the types of warnings that can be provided to a driver through a
processing and/or display unit in the cab.
 The sensor units located on the back of the vehicle can consist of different types of
detection technology, such as radar or sonar.
 Ultrasonic technology or sonar (Sound Navigation And Ranging) determines the range
of objects by emitting a transmitter pulse of ultrasonic energy.
o The resultant echo is detected by a receiver as it is reflected from the detected
object.
o The emitter is a membrane that transforms mechanical energy into a chirp
(inaudible sound wave) and sends this sound out toward the target area.
o When the sound encounters an object, it is reflected back to the receiver circuit
that is tuned to the frequency of the emitter, which then transfers the data to a
driver display unit.
Radar (Radio Detection and Ranging) technology is also used for rear object detection
systems. Radar typically operates in the ultra-high-frequency or microwave range of the
radio- frequency spectrum.
o These radio frequency waves are transmitted from the vehicle at defined intervals
within a specific coverage area.
o The sensor collects echoes from electromagnetic waves that bounce off objects
behind the vehicle. These echoes are sent to a signal processing unit and
communicated to a driver interface.
o Some processing units utilize algorithms for object detection, object tracking, and
angle measurement to provide specific distance information.
Rear object detection systems provide audible and/or visual alerts to warn drivers when
objects are detected. Some systems indicate the vehicle's distance from a detected object.
For example, the driver interface may consist of a graphical or digital visual display that
shows the distance from the vehicle to a specific object. Other visual alerts could consist
of a series of lights which change color or light up as objects are detected. These visual
alerts can be used in combination with audible alerts that vary in tone and frequency as
the vehicle moves closer to an object.
Rear object detection systems can be activated manually when needed or automatically
for continuous operation. Some rear object detection systems are connected directly to the
 vehicle's backup lights, which activate automatically when the vehicle is shifted into
reverse. Other systems are activated when the key is put into the ignition or the vehicle is
put into operation. When the systems are activated, their operation is "hands-free," and
the driver can focus on safely operating the vehicle.

TEXT / REFERENCE BOOKS

1. Powloski.J - “Vehicle Body Engineering” - Business books limited, London - 1969.
2. [Link] - “Automotive Electronics Handbook” - Second edition- McGraw-Hill Inc., -
1999.
3. Johnson, W., and Mamalis, A.G., "Crashworthiness of Vehicles, MEP, London, 1995
4. Bosch - “Automotive Handbook” - 5th edition - SAE publication - 2000.

Further Reading:
1. George A. Peters and Barbara J. Peters, Automotive Vehicle Safety, Taylor & Francis,
2003.
2. G.S. Daehn, Sustainable design and manufacture of lightweight vehicle structures, in
Alternative Fuels and Advanced Vehicle Technologies for Improved Environmental
Performance, 2014. [Link]
3. Paul M. Leonardi, Car Crashes without Cars, THE MIT Press, Massachusetts Institute of
Technology, 2012.
4. Ulrich Seiffert and Lothar Wech, Automotive Safety Handbook, Second Edition, Society
of Automotive Engineers, SAE International, 2003.
5. Ing. Konrad Reif Ed, Fundamentals of Automotive and Engine Technology, Bosh
Professional Automotive Information, Springer Vieweg, 2014.
6. Hermann Winner, Stephan Hakuli, Felix Lotz, Christina Singer Eds., Handbook of Driver
Assistance Systems, Springer Reference, 2016.
7. Ulrich Seiffert, Mark Gonter, Integrated Automotive Safety Handbook, SAE
International, 2014. doi:10.427/R-407.