Sonki Prasetya et al., International Journal of Emerging Trends in Engineering Research, 8(8), August 2020, 4768– 4772 4768 ABSTRACT A braking system is an essential value in a vehicle particularly for a safety precaution. The higher number rate of traffic accidents mostly in Indonesia shows that the dominant cause of accidents is due to the human factor. The physical condition such as tired or sleepy during driving is the primary problem according to the survey. In order to assist a person when driving a vehicle, an artificial intelligence method is necessary to be integrated to ensure the safety of surrounding people inside and outside the vehicle. This study focuses on implementing the method of identifying object and distance to provide an indicator for braking action. Images from a stereo camera is processed by a neural network technique via a mini computer to classify as well as the distance of objects. Furthermore, selection of priorities are done to obtain the intensity of braking action. The result shows that process of classification and measurement requires period around 200 ms. Furthermore, braking action done by fuzzy controller sub-system shows that the intensity has smoother signal with the object distance variation compare to the direct method. Objects are firstly identified by the presence of a stereo camera, later on the decision of braking intensity is generated by two processing unit namely conventional and fuzzy unit. This is achieved by processing the data saved from the object detection using two system via MATLAB software. The object identification result, distance measurement and the period of object detection is presented. Moreover, the response of braking intensity using data is processed with both conventional and fuzzy unit systems are also presented. he implementation of this study is for the heavy vehicle such buses or trucks that requires higher safety during the journey. Key words : Braking system, Fuzzy unit, Neural network, Stereo camera, MATLAB, Artificial intelligence camera 1. INTRODUCTION A vehicle used for transporting human from one to another place has been developed rapidly during decades. Commercial 4 wheels vehicle data show that over 80 million units are sold in 2018 [1]. Naturally, there are effects due to the higher number of owned vehicles. One of the negative impact of vehicle utilizations is the accident. The World Health Organization recorded that more than 1.3 million people died due to the vehicle accident annually [2]. In a capital city namely Jakarta, a highway accident data show that most of accident caused by human such as tired condition like sleepy [3]. Therefore, a device to assist a person when driving is necessary to avoid this problem. Moreover, this intelligent braking assistance for driver can be developed as the foundation for an autonomous vehicle. The commercial braking system for vehicles has several types. Friction is the most applied one. Hydraulic actuators normally found in light weighted vehicles for creating friction movement mechanically from the pushed brake pedal. Furthermore, for heavy weighted vehicles use pneumatic actuators. This system is a part of study of utilizing the electrical signal to move the braking actuator. There has been researches for advanced vehicles harnessing an intelligent methods. Computational perception using visual and laser scanner is developed by Martin [4]. Furthermore, braking system by vision using Support Vector Machine is done by Wang [5]. The low cost intelligent system for vehicle was the objective of Heimberger to develop three dimensional reconstruction by vision for auto parking [6]. The aim of this study is to generate an electrical signal based on the intensity for braking activation. Using two indicators namely object classification and distance, the intensity of braking is decided. The result is then transferred to the signal conditioning for the input of the actuators. 2. METHODOLOGY This paper covers two main stages of the system. Those two stages are described as the object detection and the braking decision. Camera Based Artificial Intelligence for A Smart Vehicle Braking System *Sonki Prasetya 1,3 *, Hasvienda M. Ridlwan 3 , Hendri DS Budiono 2 , Ario Sunar Bhaskoro 2 , Agung Shamsuddin 2 , Mohammad Adhitya 2 , Danardono A Sumarsono 3 1 Student of Mechanical Engineering Department, Engineering Faculty of Universitas Indonesia, Depok 16424, Indonesia 2 Mechanical Engineering Department, Engineering Faculty of Universitas Indonesia, Depok 16424, Indonesia 3 Mechanical Engineering Politeknik Negeri Jakarta, Depok 16424, Indonesia sonki.prasetya@mesin.pnj.ac.id ISSN 2347 - 3983 Volume 8. No. 8, August 2020 International Journal of Emerging Trends in Engineering Research Available Online at http://www.warse.org/IJETER/static/pdf/file/ijeter113882020.pdf https://doi.org/10.30534/ijeter/2020/113882020