Edge Artificial Intelligence Used to Activate Indicator Lights on Bicycle Helmets

The city of Bogota is characterized by the different problems it has related to traffic, especially with the traffic jams that occur within the capital. For this reason, people have opted for the use of alternative means of transportation such as bicycles, however, the number of accidents involving these road actors has increased over time. For this reason, we proposed the use of a learning algorithm, which is able to identify certain movements made by the cyclist with his head and depending on these, the directional signals of the bicycle will be activated, in order to provide information to the other entities of the actions performed by the cyclist. Classifiers refer to the process of training on a labeled data set (dataframe) to assign data points to classes for classification. In order to achieve this, classification models require a large amount of data to be able to predict more accurately. For this reason, the first thing that was done was the creation of the dataset, which was created from the data collection by means of the IMU of the microcontroller of the movements to be classified. After that, the windows of each movement were created, which for this project were 12 samples. The variance was calculated for each window and this was used to test the different classical classifiers (k-near neighbors, logistic regresión, decision trees, random forest, Naive Bayes, support vector machines and support vector machines). The best model were select using the accuracy meusurement and was easier to implement to predict the movement, which would be the logistic regression classifier. From this, the equation of this algorithm was implemented in the Arduino BLE sense 33 and this microcontroller would be placed in the cyclist’s helmet to identify the movements.