IndyCar
Real-time anomaly detection from edge to HPC-cloud in collaboration with Indycar and Intel.
Introduction
With the growth of the Internet of Things, we see an increase in importance of the analysis of data from the edge with often the results needed in real time. Indy Car series is one of the top racing series in North America. All cars are equipped with multiple cameras. The video streams captured by these cameras can be used for detection and predictive tasks to increase race safety, develop better strategies for the teams to win the race. Furthermore, sensor's data can be used as the telemetry data to provide better analysis and predictions for the drivers and the teams.
In video image analytics, anomaly detection and prediction tasks start with object detection. We create new datasets and compare three different Single Shot Multibox Detector models from TensorFlow Detection Model Zoo. We run experiments on CPU and GPU, and test mobile detection models on an Edge TPU, which is Google Coral Dev Board. Our initial results show that the Edge TPU gives the best inference time, and it is more suitable for a real-time machine learning task.
Objectives
Equipment Malfunction Prediction and Alerts
Anomaly prediction can help engineers identify issues in the hardware before they become serious. Especially at the speed at which these cars travel, small issue can quickly snowball to become catastrophic.
Recommendations with Evolving Driving Conditions
Track conditions are ever evolving, from humidity, ambient temperature and track temperature. All of these factors affect the way on-board systems perform in different ways. It can be hard for a human to take appropriate action at the appropriate time with such gradual change.
Prototype
Performance Tuning During Practice
Performance of hardware can be tuned to minute degrees based on insights provided by AI systems which can go beyond the level of pattern recognition capable by humans.