I have created a pedestrian safety framework for autonomous driving vehicles using Computer Vision (CV).
Components Used: 1.Raspberry pi 3B+ - installed with OpenCv and its associated libraries. 2.Arduino UNO (2 nos) 3.MCP2515 CAN Controller 4.LoRa WAN module 5.WebCam
I have detected pedestrians using HOG descriptor fed into a pretrained Support Vector Machine algorithm in OpenCv.
In order to send a signal to the ECU of the automobile after pedetrian detection i used a CAN controller(MCP2515) which converts the oputput from the raspberrypi into a signal which could be transmitted through CAN BUS.
In order to indicate the vehicle which is having a blind spot i used LoRa WAN interfaced with an arduino at receiver end.