Content uploaded by Damian Schofield
Author content
All content in this area was uploaded by Damian Schofield
Content may be subject to copyright.
1
Forensic Animation: Measuring the Reliability and Accuracy of
Computer Generated Animation Used in the Courtroom
Dr M Tarek Shalaby, Ms. N Hussin, and Dr D Schofield
The University of Nottingham
Abstract
This paper discusses the findings of ongoing research
into forensic animation at the University of Nottingham.
The paper has six sections. The Introduction part explains
the general context about the use of forensic animation
for court litigation. The second part describes challenges
for the computer animation in the courtroom. The third
section deals with procedures used to develop a
particular forensic animation case study. It depicts a
fatal road accident involving a car and two motorbikes on
a dual carriage-way in the UK. The fourth section
discusses the development of a new methodology for
measuring the reliability and accuracy of forensic
animation. The fifth section elucidates analysis on
knowledge theory and deductive reasoning. Finally, the
conclusion part focuses on demonstrating the extent to
which a particular frame of animation carries reliable
and accurate information (evidence) which will help a
judge and jury to understand complex events.
Keywords: Forensic animation, accident reconstruction,
syllogism, virtual reality.
1: Introduction
Recent and rapid developments in PC technology and
the huge potential market for desktop Virtual Reality
(VR) have created a climate where novel applications
have emerged. The computer games market has driven the
development of software tools for the creation of 3D
environments alongside specialist 3D graphics accelerator
boards and input/output (I/O) peripherals for PC games
systems. Whilst much of the development is for the
leisure industry, there are many real industrial
applications being developed under rigorous guidelines
[1].
Forensic animation involves the use of animated
computer graphics to recreate an event such as an
automobile accident, the collapse of a building, an assault,
or the workings of a mechanical device, from a variety of
perspectives [2].
In a legal context, evidence is information by which
facts tend to be proved, and the law of evidence is that
body of legal rules regulating the means by which facts
may be proved in courts of law and tribunals and
arbitration in which the strict rules of evidence apply [3].
Computers may be used to illustrate evidence, but it
does not necessarily make the evidence more accurate or
reliable. To create an animation, data has to be entered
into the computer. An animation is, therefore, only as
good as the information upon which it is based. There is
no substitute for meticulous investigation and careful
analysis.
2: The challenges for computer animation in
the courtroom
A visual image can have a very strong impact on a
jury. Such images can be easily manipulated, and the
potential for misleading a jury is ever present. A good
example of this (in a situation that is analogous to a
computer simulation) is Gladhill v. General Motors. In
this case, a videotaped demonstration (by the defense) of
the braking characteristics of a 1980 Chevrolet Citation
was admitted into evidence. The issue was that the
Plaintiff's accident occurred at night, on a sharp downhill
curve, and the Defense's demonstration was done in the
daytime, on level ground, and was conducted by an
experienced test driver. The defense tried to argue that the
test was not a reconstruction of the accident, but rather “a
demonstration of certain operating characteristics of the
vehicle in question”. The court of appeal did not agree
with this argument and held that such evidence was
misleading: It is easy to understand why the jury might be
unable to visualise the plaintiffs' version of events after
watching this video. Indeed, the circumstances of the
accident, as alleged, were so different from the
Proceedings of the Seventh International Conference on Information Visualization (IV’03)
1093-9547/03 $17.00 © 2003 IEEE
2
demonstration as to make the results largely irrelevant if
not misleading [4].
There is a great deal of scope for tampering with the
evidence in computer-generated displays. This possibility
was recognised in the dissenting judgement of Justice Van
Graafeiland in the US case of Perma Research &
Development v Singer [5] . The learned judge stated that
although a computer has tremendous potential for
generating more meaningful evidence, “it presents a real
danger of being the vehicle for introducing erroneous,
misleading or unreliable evidence.”
Even where there is no deliberate attempt to tamper
with the evidence, computer-generated displays can be
unintentionally misleading. Computer animations in
particular rely a great deal upon data collection, human
judgement and speculation at each step of the animation
process.
In view of the potential for misleading the jury and
tampering with the evidence, it is evident that the most
vital issue in animation is the reliability and accuracy of
the information that may be used to create the animation.
Two questions that may be regarded as essential in
this respect are (1) the objective of the animation, e.g. the
police investigator instructed the animator to animate the
evidence demonstrating that the car had made the turn at
the junction, hence, the sequence of animation shows that
the car driver could not see the motorbike coming from
the opposite direction; and (2) what is the critical issue(s)
from the judge and jurors’ point of view, e.g. the trial may
look at various traffic offences with regard to the
accident.
Several courts have voiced specific criteria that an
animation must meet before it can be admitted as
evidence :–
x The animation must be a fair and accurate
representation of the evidence to which it relates;
x It must be relevant; and
x Its probative value must substantially outweigh
the danger of unfair prejudice, confusion of the
issues, or misleading the jury. [6]
A computer animation is predominantly used only as
a presentation tool; it is not often used as an analytical
tool. An animation is usually created based on
information provided by a witness or data provided by an
expert [7].
3: A forensic animation case study
This case study refers to a fatal accident, which took
place on a dual carriageway in the UK. It involved two
motorbikes and a car. The car was making a turn into a
junction and the two motorbikes were approaching from
the opposite direction. At the mid-point of the turn, the
motorbikes collided with the car and both motorcyclists
were killed.
The data acquisition process in this case involved
obtaining evidence from :–
x Police accident reports.
x Original police photographs.
x Witness’ statements.
x Police drawings (e.g. Figure 1).
x Crash investigation reports (e.g. Figure 2).
Figure 1: Police plan illustrating the layout of the accident
scene
Figure 2: Crash investigation report describing the
calculation of the coefficient of friction between the vehicle
tyres and road surface
The first task undertaken was the development of
the three-dimensional computer model of the immediate
environs. The model was developed based on the
ordinance survey road layout plan of the scene and police
survey data. The features in the model were accurately
positioned using the available data (from the geometry of
the road layout to the positions of street furniture such as
lamp posts and bus stops).
Three-dimensional models were created to represent
the vehicles in the scene including the car and the two
motorbikes involved in the accident. The three-
dimensional vehicle models used were created such that
the dimensions accurately matched the measurements
specified in the accident reports.
Proceedings of the Seventh International Conference on Information Visualization (IV’03)
1093-9547/03 $17.00 © 2003 IEEE
3
A number of calculations were undertaken to allow
the vehicles to be accurately animated within the virtual
world. The animated vehicle positions are based on the
results of police calculations and further calculations were
undertaken to correlate the police results to a series of
small time increments of between 0.1 of a second and 1.0
second. Calculations were performed at each time interval
to allow the animator accurately position a vehicle within
a template at that particular moment of time. For example,
as the motorbikes approach the junction, available data is
limited and the police accident investigator assumed a
constant velocity, hence the animator used an interval of
1.0 second for the calculation templates. However, when
the motorbikes begin to skid, more accurate data on the
calculation is available, and the event happens very
quickly, so the interval used during this time period was
0.1 of a second.
Figure 3: Still from a computer generated animation
demonstrating the point where the car is making the turning
Figure 4: Still from a computer generated animation
demonstrating the movement of the two motorbikes
The rendered images and animations were then added
to a browser based presentation system, allowing easy
access and recall of all of the information created.
Multiple views of the virtual world were created, these
consisted of both static and animated viewpoints (from
overhead views generated for clarity to views
demonstrating witness viewpoints). Rendered frames
from the animations reduced are shown in Figures 3 and
4.
During the production process, steps must be taken to
ensure accuracy and reliability. The animator and expert
witness who assisted in directing the animation must be
ready to testify:-
x that the underlying data are accurate,
x that the process by which the data were fed into
the computer provides reasonable assurance that
error was avoided, and
x that tests were used to maintain the accuracy and
reliability of the hardware and software
employed. [7]
Items that fall under the category of collision
evidence include positions of rest, tire marks, roadway
markings, damage to vehicles, and damage to property.
Wherever possible, a technique that takes into
account both accident scene information and vehicle
damage information should be used to perform the
collision reconstruction.
In the mathematical calculations in this case study,
prepared by a senior crash investigator, the calculation of
the coefficient of friction between the tyres and road
surface utilised a trajectory-based technique, making use
of the Law of Physics principle. Hence, the forensic
animator had used the result of the calculation from the
crash investigation to animate the accident.
4: The proposed approach
Epistemology is the branch of philosophy that studies
knowledge. It attempts to answer the basic question: what
distinguishes true (adequate) knowledge from false
(inadequate) knowledge? [8].
The theory of knowledge seems to be an effective
way to measure the reliability and accuracy of computer
animation as a way of presenting evidence in the
courtroom.
Wisdom (e.g. perception in the judgement)
Knowledge (e.g. admissibility, collision
reconstruction)
Information (e.g. police statements, witness’
statements)
Data (e.g. photographs, numerical measurements)
Rules and formation (e.g. data acquisition)
Symbols (e.g. scientific formula)
Figure 5: Knowledge hierarchy explaining the evolution
from “symbols” to “wisdom”
Proceedings of the Seventh International Conference on Information Visualization (IV’03)
1093-9547/03 $17.00 © 2003 IEEE
4
Figure 5 illustrate the knowledge hierarchy from the
“symbols” through to “wisdom”. Data means letters,
numbers, lines, graphs and symbols, etc. – that are used to
represent events and their state organised according to
formal rules and conventions. Examples of data may be
represented by numerical measurements, photographs of
vehicles and photographs of the accident scene.
Information means the cognitive state of awareness
(as being informed) given representation in physical form
(data). This physical representation facilitates the process
of knowing. Examples of information may include
references to the chain of events, which occurred as stated
in the witness’ statements, police statements, and the post-
mortem report.
Knowledge means the cognitive state beyond
awareness. Knowledge implies an active involvement and
understanding and the ability to extend the level of
understanding to meet life’s contingencies. Examples of
knowledge may refer to the concepts in developing the
animation such as, collision reconstruction, demonstrative
evidence, expert witness testimony and admissibility.
Wisdom implies the application of knowledge as
contained in human judgment centered on certain criteria
or values that are generally accepted by the culture of
society. Wisdom is the point where judgment and verdict
shall take place. The perception (wisdom) of judge and
jury in analysing how information (e.g. the animation) can
prove knowledge (e.g. the concepts) by using data (e.g.
the substantive evidence).
In applying this approach, this paper embraces a
deductive reasoning pattern. Deductive reasoning is a part
of human thought process, often categorised under
Human Computer Interaction (HCI). Deductive reasoning
works from the more general to the more specific. Figure
6 illustrates the pattern of analysis in deductive reasoning.
Theory:
A set of ideas formulated to explain something.
General hhypotheses:
Supposition or conjecture put forth in the
form of a prediction according to a theory,
observation, problem.
Figure 6: Deductive is reasoning from the general to the
particular
A deductive argument offers two or more assertions
that lead automatically to a conclusion. In the research
undertaken at Nottingham, which is described in this
paper, deductive analysis has been used to examine the
presence of evidence in a particular animation. Then, if
the evidence is reflected in the animation, the researchers
have attempted to measure the extent to which the
animation is reliable and accurate, based on the evidence.
The following is an example of a sound deductive
syllogism:
Premise: All birds have wings.
Premise: A parrot is a bird,
Conclusion: A parrot has wings. [9]
Figure 7, illustrates the pattern, based on the rules of
deductive reasoning which have been used in measuring
the reliability and accuracy of the computer-generated
animation against evidence, concepts and issues.
HUMAN COMPUTER INTERACTION
Deductive reasoning
FORENSIC THEORY OF
ANIMATION KNOWLEDGE
Evidence
E.g. Crash Investigation Report
Police Report, Witness’ Statement
Concepts
Collision Reconstruction
Expert Witness
Demonstrative Evidence
Admissibility
Figure 7: Link between Forensic Animation, Theory of
Knowledge and Human Computer Interaction
The measurement is concerned with the analysis as to
whether animation and simulation may or may not help to
explain complex events.
As mentioned in the earlier part of this paper, most
studies of knowledge attempt to answer the question of:
what distinguishes true (adequate) knowledge from false
(inadequate) knowledge? What is important in examining
the reliability and accuracy of an animation, is to
investigate which evidence has been used in developing
the particular animations.
As presented in court, the animation should carry
accurate and reliable information from the evidence. The
judge and jury will view the animation and ask the expert
witness questions such as :–
Proceedings of the Seventh International Conference on Information Visualization (IV’03)
1093-9547/03 $17.00 © 2003 IEEE
5
x How did you determine at what particular time
the car made the turn ?
x Why did you assume that the surface of the road
is viewed in such manner ?
5: Analysis using theory of knowledge
In analysing this issue, two main areas from theory of
knowledge can be taken into account :–
x Types of knowledge.
x Conditions for knowledge.
5.1: Types of Knowledge
The first type of knowledge is competence. An
example of competence is when an individual displays
competence, the interpretation is, that he or she knows
how.
The second type is acquaintance. An example of
acquaintance is when an individual may be said to know
that with which he or she is acquainted. To say that one
knows something in this sense is to say that they have had
some experience with what they know.
The third type is recognition of information as being
correct – this is knowledge in the (correct) "information"
sense. To know is to recognise correct information as
being correct. Example: I know that 2 + 2 = 4 because I
possess the information that 2 + 2 = 4, the information is
correct, I consider it to be correct, and I have a good idea
why I think it is correct.
The following is a witness statement from our case
study:
Mrs. R: Remember? Well I couldn’t actually see them
hitting me because I’m on the driver’s side and
they are partly shielded by the window line on
my nearside from the photographs that I’ve
seen in the paper.
This witness statement is illustrative of the different types
of knowledge mentioned above.
Competence – The statement, “they are partly
shielded by the window line on my nearside from the
photographs that I’ve seen in the paper”, demonstrates
competent knowledge on the part of Mrs. R, based on
what she saw in the paper.
Acquaintance – This is demonstrated in the
statement, “Well I couldn’t actually see them hitting me
because I’m on the driver’s side.”
There is an association with competence here, in that
there is knowledge (“the photographs that I’ve seen in the
paper”) in cases where the individual does not have the
ability to recognise the thing (I couldn’t actually see them
hitting me because I’m on the driver’s side).
Recognition of information as being correct – Mere
possession of information (“the photographs that I’ve
seen in the paper”) is not sufficient to be considered
correct knowledge. In her statement, Mrs. R may have
regard the photographs she saw in the paper as conveying
correct information, but the accident investigation still has
to determine the correctness of her statement.
5.2: Conditions for knowledge
There are three conditions for knowledge. The first
condition is truth. The second condition is acceptance and
the third one is justification [10].
The example below demonstrates how the conditions
for knowledge apply to another witness statement from
our case study
Mrs. R : At the junction with J Road, I pulled into the
protected area and waited while two or three
other cars passed in the opposite direction.
Truth – “J Road” is true if and only if J Road exists.
In this case “J Road” exists, hence, it is true in her
statement.
Acceptance – Mrs. R’s statement, “I pulled into the
protected area and waited while two or three other cars
passed in the opposite direction”, may be accepted as
being her version of events, but may not necessarily be
endorsed as being what actually took place.
Justification – Whether or not Mrs. R had actually
pulled into the protected area carries a level of
justification between reasonableness and complete
certainty, for the accident investigator and animator.
By taking these two main areas, (i.e. the types of and
condition for knowledge) as the main components in this
measurement of accuracy and reliability, an analysis can
be done by cross-tabbing the animation with substantive
evidence and concepts illustrated in Figure 7. This
syllogism pattern could be applied as a useful tool for
examining the type of knowledge and the conditions for
knowledge.
5.3: Deductive Reasoning
Further details on the analysis using deductive
reasoning can be seen in Figure 8 below, which explains:-
that the information from the crash investigation (CI) has
been used in the computer generated animation (CGA),
based on the collision reconstruction (CR), therefore,
aspects of the animation which are assessed in this
manner can be certified to be reliable and accurate.
Proceedings of the Seventh International Conference on Information Visualization (IV’03)
1093-9547/03 $17.00 © 2003 IEEE
6
CGA CI CGA
CI CR CI
CR
C1=CGA CR=C1, therefore CR=CGA
Figure 8: Relationship between crash investigation (CI),
computer generated animation (CGA), and collision
reconstruction (CR)
6: Conclusion
The reliability and accuracy of a forensic animation
depends on the precision and verifiability of the data used
to create it. In the road accident case study, for example, it
is important to know prior-to-impact data including the
speeds of the two vehicles, the precise directions in which
they were traveling, the masses of each vehicle, the points
on each vehicle at which contact was first made, and
whether or not (and to what extent) either vehicle was
accelerating or decelerating. This data might be gathered
from the testimony of one or both drivers, eyewitnesses,
police officers who came to the scene after the accident,
or engineers who analysed the structural damage to the
vehicles and the skid marks (if any) left on the roadway.
The nature and condition of the pavement, the weather,
the condition of each vehicle's tires, and the reaction times
of the drivers could provide important additional data.
Even a slight change in one of the parameters and
concepts can result in a drastic change in the legal
judgment.
The effectiveness of the epistemological approach
using types of knowledge, conditions for knowledge and
the deductive reasoning can be seen in the paper. The
analysis of the types of knowledge distinguishes the type
of information itself when looking at a particular
statement. The conditions for knowledge can be used to
make a specific analysis to determine the truth,
acceptance and justification values in the witness
statement. The authors believe that the deductive
reasoning patterns described are a sound methods for
investigating as to whether the animation reflects reliable
and accurate information from the evidence used.
7: References
1
SCHOFIELD, D., DENBY, B, AND HOLLANDS, R. Mine
Safety in the Twenty-First Century: The Application of
Computer Graphics and Virtual Reality, in: Karmis, M., ed.
Mine Health and Safety Management, Colorado: Society of
Mining, Metallurgy, and Exploration, Inc., 2001, pp. 153-174.
2 TECHTARGET, Forensic Animation - A Whatis Definition
[online]. TechTarget.com. URL:
<http://whatis.techtarget.com/definition/0,,sid9_gci841381,00.ht
ml> [06 March 2003].
3 KEANE, A. The Modern Law of Evidence. 5th Ed.
Butterworths. 2000. p.1
4 THOMASON, K. L., Using Computer Generated Evidence, in
Articles and Cases Related to Demonstrative Evidence [online].
Arizona: Legal Imaging, 1994-1996, URL:
<http://www.legalimaging.com/evidence.html> [6 March 2003].
5 Perma Research & Development v Singer. 542 F 2d 111 (2d
Cir) cert denied, 429 US 987 (1974).
6 See Harris v. State, 13 P.2d 489, 494 (Ok. 2000); Clark, 529
S.E.2d at 536, cited in WILLIAMSON, THOMAS W. Jr.,
Admitting Into Evidence Computer Animations and Simulations
[online]. Richmond, Va: Williamson & Lavecchia, L.C. URL:
<http://www.wllc.com/Library/Copy_of_Admitting_Into_Evide
nce_Computer_Graphics_and_Simulations.htm> [6 March
2003].
7 3-Dimensional Computer Animation in the Courtroom (White
Paper) [online]. Redmond, WA: Advanced Solutions, LLC, July
25, 1995, URL: <http://www.halcyon.com/asllc/cle_ani5.html>
[6 March 2003].
8 HEYLIGHEN, F. Epistemology, Introduction, in:
HEYLIGHEN, F., JOSLYN, C. & TURCHIN, V. (eds):
Principia Cybernetica Web [online]. Brussels: Principia
Cybernetica, URL:
<http://pespmc1.vub.ac.be/REFERPCP.html> [6 March 2003].
9 AUSTIN, M. Deductive and Inductive Arguments, [online]
URL:
<http://webpages.shepherd.edu/maustin/rhetoric/deductiv.htm>
[6 March 2003].
10 MATTEY, G. J. Lecture Notes, Lehrer's Theory of
Knowledge, second edition, Chapter 1: The Analysis of
Knowledge [online]. UC Davis Philosophy Department, URL: <
http://www-philosophy.ucdavis.edu/phi102/tkch1.htm>
Proceedings of the Seventh International Conference on Information Visualization (IV’03)
1093-9547/03 $17.00 © 2003 IEEE