Using Artificial Intelligence to Determine the Speed Change of Frontal Collisions
2022-01-0998
03/29/2022
- Event
- Content
- The speed change (Delta-V) sustained by a vehicle during a collision is a key element in many traffic collision investigations. Accident reconstructionists have a variety of investigation techniques that can be used to estimate speed change including crush energy, momentum based simulations and event data recorders. This paper proposes that artificial intelligence can be used to analyze photos of damaged vehicles in order to estimate the speed change. An artificial intelligence model was trained using images from over 1000 investigations of frontal collisions (the training set). The resulting model was tested using images from a new set of 250 frontal collision investigations (the test set). In about 85% of the test set, the resulting model predicted the speed change to within ±5mph of the original collision investigation results. By adopting artificial intelligence based algorithms, the accident reconstruction community can gain an additional check on their speed change calculations. The automotive insurance industry may also benefit from being able to gain speed change estimates directly from damage images.
- Citation
- King, M., "Using Artificial Intelligence to Determine the Speed Change of Frontal Collisions," SAE Technical Paper 2022-01-0998, 2022, .