Autonomous Robot Control Using Night Vision Camera and TOF Sensor for Depth Scanning and Real-Time Object Recognition

Abstract

Autonomous Land Vehicle (ALV) isstronglydependent on correct sensormeasurements. Thus, in
thisworkweprovide a solution for night drivingusing a preciseobject classification detector based on
TensorFlow: trafficsignalization, cars and pedestrians. A neural network wastrainedusing night camera
images and used for trafficsignals recognition and automatedvehicle control. Wealsoinclude a
depthsensorinfraredsensor to obtain the correct distance to a detectedobjectproviding a solider setup in this
type of application as this system isalso capable of not onlydetecting the type of object but also the distance
to it. This system works in real-time and is capable of autonomousdriving. Wepresent the resultsfrom the
neural network accuracy, the complete system description and the code used by the motor drivers to
control the robot based on sensors output and object recognition