DESIGN OF AUTONOMOUS BUS USING MACHINE LEARNING AND LYAPUNOV STABILITY CONTROL
According to the Regional Development Study Association study, the number of motorized vehicles in Indonesia has reached 138.5 million units at the beginning of 2018 with 87% of them being private vehicles, which caused congestion with an average time wasted on roads by 55 hours per year with 107,96...
Saved in:
| Main Author: | |
|---|---|
| Format: | Final Project |
| Language: | Indonesia |
| Online Access: | https://digilib.itb.ac.id/gdl/view/43254 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Institut Teknologi Bandung |
| Language: | Indonesia |
| Summary: | According to the Regional Development Study Association study, the number of motorized vehicles in Indonesia has reached 138.5 million units at the beginning of 2018 with 87% of them being private vehicles, which caused congestion with an average time wasted on roads by 55 hours per year with 107,968 accidents in 2018. This was caused human errors of drivers, such as inability to drive or non-compliance in following traffic signs. One of the solutions to this problem is to increase the number of bus transportation services with autonomous system. To reduce the level of accidents, the autonomous system has an object detection feature that is capable of detecting objects in front of the vehicle and performing control responses quickly and precisely.
This research is to design an autonomous bus prototype that is able to avoid object by changing it lanes and stopping itself. The object detection system is designed using a camera that is connected to a computer. The image data is then processed with machine learning using a convolutional neural network-based Single Shot Multibox Detector (SSD) algorithm that is able to locate object in an image and classify the detected object. Autonomous control is designed using the Lyapunov stability control algorithm so that the vehicle is able to follow the designated path. The design results were tested by simulation along with the implementation of prototypes using Lego EV3 that modified so that it can represent a bus.
This research produced an object avoidance system that able to generate a control signal according to the detected object with 90.47% success rates with Lyapunov stability control achieve mean average error 2.3 cm from setpoint on maximum speed of 5,33 cm/s.
|
|---|