‘PACE-AI’
SENTINEL will embed the PACE-AI, the PedestriAn Collision forensics Evaluator from Coventry University to compute the collision impact parameters, like the vehicle impact speed. As impact speed and injury severity are linked, SENTINEL will seamlessly transfer the PACE-AI predictions into its brain trauma AI model for in-situ triage predictions.
PACE-AI: The Pedestrian Collision Forensics Evaluator from Coventry University | Coventry University
PACE-AI requires an estimate of the bumper and windscreen damage locations, the pedestrian height and weight, a search tolerance for the head impact strike on the windscreen and… that is it. PACE-AI will calculate, in seconds, all the plausible combinations that can lead to this collision and will provide to the forensics’ team the vehicle impact speed, pedestrian crossing speed, and pedestrian.
1. What is PACE-AI and what data does it require?
PACE-AI is a forensics support system that aims to reduce pedestrian collision design space in seconds using only a few inputs (vehicle profile, damage coordinates, and pedestrian height and weight) and no advanced forensic calculation skills. It is not a replacement for forensic reconstruction analysis, but a step towards narrowing the design space instantaneously by providing realistic accident characteristics to assist effective accident reconstruction. It can be used by the roadside.
2. How is PACE-AI developed?
PACE-AI is trained using over 3000 Madymo[i] pedestrian computed collisions. Madymo is a multibody solver, which was chosen thanks to its capability, evidenced in research publications, to predict accurately pedestrian kinematics. PACE-AI, via a patented search and convergence algorithm, extracts from vehicle profile and damage, as well as pedestrian anthropometry, plausible vehicle impact speed, pedestrian crossing speed, crossing direction and gait, in this in seconds.
3. What is the scientific underpinning PACE-AI?
Shrinivas, V., Bastien, C., Davies, H., Daneshkhah, A., Hardwicke, J., Neal-Sturgess, C. E., & Lamaj, A. (2024). Integrating Machine Learning in Pedestrian Forensics: A Comprehensive Tool for Analysing Pedestrian Collisions. SAE Technical Papers, Article 2024-01-2468. https://doi.org/10.4271/2024-01-2468
4. How can PACE-AI help me with forensic investigation?
Suppose you are investigating a pedestrian collision, which could be a hit-and-run case. Evidence on the road, necessary to perform any accident forensic analysis and accident reconstruction, can be sparse and take a long time to collect, i.e., the point of collision may not be evident, and the weather conditions not favourable to collect this relevant data. You would like to calculate, in seconds, by the side of the road, what happened at the time of the collision. That is where PACE-AI comes in. PACE-AI is a simple web-based application that can calculate, from the bumper and windscreen damage and the estimated height and weight of the victim, the most plausible set of circumstances at the time of impact. PACE-AI will help you narrow the range of possible scenarios and have a more realistic idea of what happened. This way, you can determine the driver’s level of responsibility, and take preventive actions, if necessary, while you wait for a more detailed and conclusive collision reconstruction report. The computation results from PACE-AI can also be provided later to the forensic investigation team, making the final reconstruction quicker, hence saving investigator resource, time and cost.
5. Is PACE-AI validated?
Yes, it is validated using the pedestrian collisions available from the Road Accident In-Depth Studies (RAIDS) database. PACE-AI’s performance was within an error margin of 5 km/h.
6. Who to contact about PACE-AI.
Enquiries about PACE-AI can be made to pedestrian-collision@coventry.ac.uk or christophe.bastien@coventry.ac.uk.
[i] Simcenter Madymo, software for simulating human safety in transport and road users, is part of Siemens PLM Software