Determining the speed of motor vehicles involved in collisions is central to virtually all cases where fatalities have occurred. I have prosecuted countless criminal driving fatality cases wherein speed was integral to determining liability. While there are cases where vehicle speed cannot be reliably determined, speed can typically be determined from measurements obtained at the scene, eyewitness testimony, assessment of motor vehicle damage (crush analysis) and downloading the on-board computers found in the involved vehicles. For all but eyewitness estimates, expert evidence is required to establish speed.
In more recent years, investigators, experts and attorneys have ventured into speed determination using video evidence. In appropriate cases, provided the analysis is properly undertaken utilizing the appropriate experts, speed (or a range of speeds) can reliably be determined from video evidence. However, in most cases, the analysis has not been appropriately conducted, and the correct experts have not been engaged in the analysis. Further, untested by effective cross examination, speed estimates based on video will appear to be far more certain than can be justified.
R. v. Botelho, 2018 ONSC 4097, a 2018 decision of the Ontario Superior Court of Justice, is a case wherein speed determination premised on video evidence was a central feature of the case. The defendant was charged with dangerous driving causing death while street racing. Several segments of video obtained from different locations were used in court to show portions of the impugned driving. It was these video segments that were used to provide expert evidence of vehicle speed.
Many cases fall into error at the outset when a video expert alone attempts to determine speed without working with a collision reconstruction expert and vice versa. That problem did not occur in this case as it was correctly recognized that two experts were required to properly assess vehicle speed based on video.
A forensic video analyst was qualified as an expert witness to provide technical evidence regarding the recovered video. He was asked to determine the frame rate of two different video segments. He found that one video had a frame rate of 14-15 frames per second while the other video had a frame rate of 15 frames per second. The expert advised the court that he was not qualified to provide a speed estimate premised on video evidence and cautioned that there were too many variables involved in speed calculation from video to reach an accurate conclusion. He further stated that simply watching the video evidence could not achieve this objective.
In providing the foregoing expert evidence, the forensic video analyst was cognizant of the limits of his expertise and did not stray beyond those limits. He knew that a collision reconstruction expert was needed to integrate the frames per second data into speed analysis.
A second expert, qualified in the field of accident reconstruction and speed determination from video, determined the speed of the victim’s vehicle at a particular point in time to be 118 km/h based on frame rate and distance travelled. In cross examination, the expert was asked to consider variations in frame rate, number of frames utilized, constancy of speed, and measurement of the observed vehicles. He agreed that his speed assessment was only for a specific point in time. He could not comment on acceleration or deceleration or whether either of the vehicles stopped at any point in their travel.
In assessing the evidence of the accident reconstruction expert, the Court said the following:
53 Cst. Mooney’s evidence and his methodology in particular, was interesting. Given the pervasiveness of cameras and videos, it struck me that the idea and the theory behind it was an interesting prototype for an analytical model that might offer speed ranges or complement other speed data. The proposed methodology holds the potential of being useful in future cases. However, in its current formulation and application the methodology produced results that were not reliable.
54 The idea that distance could be measured in terms of framesper second sounded good in theory. But the determination of the number of total number of framesneeded to go from point ‘A’ to point ‘B’ assumed the ability to pinpoint with precision the correct framefrom which to begin the counting. That pinpointing however is vulnerable to the frailties of perception. One could be off by a number of framesin the identification of that pinpoint depending on the variations in the camera angles, parallax factors, and one’s own perspective. A variation of even a few framesresults in vastly different estimates.
55 This became evident in the number of different scenarios that Cst. Mooney was asked to consider and calculate. Even with the use of generous margins of error, the variability in possible results of about 40 km/hour could not allow the trier of fact to come to a conclusion of any sort beyond a reasonable doubt; the doubt would fall within the 40 km/ hour range. Such a variation is especially significant this case because the lower estimates would support a conclusion that the drivers were merely speeding above the permitted speed limit. The higher estimates would likely place them in the zone of dangerous driving. As between those options, it would be very difficult, absent other evidence, to come to any conclusions beyond a reasonable doubt about the actual speeds of the vehicles in question. In other words, at this stage, the model of analysis has built within it reasonable doubt.
56 Apart from this very significant technical failure, Cst. Mooney’s agreement that his analysis was focused on only a very small part of the route placed very serious limits on the utility of his evidence even if it were reliable. He had no way of knowing if the vehicles accelerated at some points, if they decelerated or stopped at other points on the route, or if their speeds varied at all. In such circumstances, even if his estimate of 118 km/ hour were held to be robust, it could only apply to a tiny fraction of the route that was analyzed.
The Court ultimately found that the expert evidence itself created reasonable doubt as to the speed of the vehicles involved in the alleged race. This was not strictly a criticism of the expert evidence itself but rather a reflection of the many variables that must be taken into account when assessing the reliability of vehicle speed as determined from video evidence.
As an aside, a police officer testified that accurate speed evidence could have been obtained had the police downloaded the data from the EDRs which both vehicles had on board. That step was not taken as the cost of doing so would have depleted the police budget. The Court found that the failure to collect this highly relevant and reliable evidence was “inexcusable,” a finding hard to dispute.
Lessons Learned
The police and prosecution were correct to use both a video expert and an accident reconstruction expert. By doing so, they avoided the most common speed determination error – relying upon one expert to do the work that calls for two areas of expertise.
The accident reconstruction expert gave a speed estimate of 118 km/h in direct examination. By the end of cross examination, the evidence was expanded to a range of 87-127 km/h. Speed estimates, by definition, are merely estimates and should be expressed as a range of speeds. Therefore, the court should be told, in direct examination, the many variables involved in assessing speed from video and what the full estimated range of speed is taking into account those variables. The failure to do so can give the impression that but for cross examination, the court would have mistakenly been led to believe that the speed estimate was more accurate than justified. It must always be remembered that expert witnesses are called to assist the court, not the party that called them to the stand. Relying upon defense counsel to draw out the true range of speed estimation gives the appearance of an expert not focused on providing full and frank assistance to the court in direct examination, an appearance that must be assiduously avoided.
Finally, denying the court the best evidence of vehicle speed (EDR data) because of resource challenges is truly inexcusable. Surely the alleged criminal death of a motorist merits a proper investigation. The goal should always be to find the truth, not just the truth we can afford.