In this day and age autonomous cars are on the rise. As we increase our understanding on artificial intelligence, improvement in radar, GPS and cameras has finally allowed REKA to present you CRETA, our very own self-driving car!
CRETA stands for Conscious REKA Transport Apparatus, a.k.a. our self-driving car. Focusing on retrofitting existing vehicles, anyone who owns a regular car can get themselves a self-driving car. Although, the project is ever growing with its many phases and looking for pledges to support the innovation, the system itself has been tested on the road and proof of concept is verified.
On the mechanical side of things, we’ve installed a gear system that is fitted between the steering wheel and the car’s boost kit. A power window motor is mounted between the steering wheel and the speedometer to control the steering of the vehicle. To get the self-driving car drive around without hitting any obstructions, we used a method called Visual SLAM, or Simultaneous Localization And Mapping. Essentially, it analyzes the environment by mapping the surroundings based on the data captured and gathered from a visual sensor—a webcam. With a little bit of calculation, the objects captured are recognized and a map is generated. As for controlling the steering on the highways, we use Hough Transform to detect and try to maintain the center of each lane. Histogram of Oriented Gradient (HOG) was also used to detect pedestrians, picking up features from people’s images to train the program. In real-time, the sensor picks up an image of a person, compare the extracted HOG features to check how close the feature is to the trained image.
In order to speed up the object detections for the camera, we perform background subtraction using the Mixture of Gaussians (MoG) method. This allows our self-driving car to discriminate between the road in front and the other cars; or just about any other object around its peripheral view. Although it’s easy to confuse the two algorithms mentioned (HOG and MoG), but just remember: one is for Gradient, while the other is for Gaussian. The final segment that made the whole thing work, is the control unit. The control unit handles the acceleration, brake, and steering control of the vehicle itself. Coordination of controlling the control unit was made based around the REKA Innovation Gear (RIG). The RIG translates every decision the CPU makes through actuators connected to the acceleration, brake and steering control of the car, making the car an autonomous vehicle. There are still a lot of possible improvement and additions that can be made to our self-driving car before it can be safely used on Malaysian roads on a daily basis.
Self driving car
CRETA demonstration at MAEPS
CRETA Nano at MAI
MISSION REPORT STILL UNAVAILABLE
INSERT ABSTRACT HERE.
REKA Innovation Gear
Use of REKA's original product, the RIG, to to connect multiple sensors for receiving and transmitting data telematically.
Installed a gear system; fitted between the steering wheel and the car’s boost kit. A power window motor is mounted between the steering wheel and the speedometer to control the steering of the vehicle.
Actuators installed to control the acceleration, braking and steering of the vehicle.
Control unit handles the acceleration, brake, and steering control of the vehicle. Coordination of is obtained from the CPU. Handling of actuators from the control unit is based on the REKA Innovation Gear, RIG. RIG translates the decision of CPU through the actuators.