Answer To: INSTRUCTIONS: 1. Each question must be clearly labelled 2. Due Date and time: 23 rd of January 2022...
Tanisha answered on Feb 09 2022
Self-driving Cars
Self-driving car act as the computer controlled car that drives itself. A complete autonomous vehicles which can be used for both the consumers and commercial fleets, as defined by the Society of Automotive Engineers(SAE), are way off now, but many of the automakers of today’s era are using the semi autonomous driving technology. Automakers such as Ford, Toyota, Waymo and Tesla build up the self driving cars with the latest technology. Self –driving cars comprised of sensors that perceive its ambience such as thermographic cameras, radar, lidar, GPS and sonar.
Engineers of self-driving cars will make use of vast amount of data from image recognition systems, using machine learning algorithm and neural networks to build up the systems. Here the neural network learns to identify all the objects like traffic lights, curbs, pedestrians and other things of the ambience. For Example: Google’s self-driving car project called Waymo, use a mixture of sensors, LIDAR, cameras and combines all the data those systems are generating to identify all the things and predict what all the objects comes in the front. Waymo sets a destination based on the passenger input, the software calculates the route, and the lidar sensor creates a 3D map of the current environment and detecting car position relative to the 3D map, collects input from Google street view and an override function which enables human to take control of the vehicle. For a self driving car, to avoid collision, it detects and classifies the traffic lights. Various CNN based transfer learning models like VGG16, DenseNet121 are on freely available Lara and Lisa traffic light datasets. These datasets can help us to segregate the images into different light classes like dayRed, dayYellow etc. .Random Forest algorithm identifies the color of the detected traffic light from the experiments and later on with CNN models will help to extract features from the ambience.
In vehicle autonomy, the SAE’s automation level defines driving mode based on different levels. We have defined five stages of automation which are as follows:
1. Level 1 comprised of cars that will have some driver assistance systems such as acceleration, lane changes and cruise control.
2. Level 2 comprised of some advanced cruise control system or autopilot systems that allows the car to take all the safety actions like emergency braking, accelerator system. In this automation, driver needs to be alert at the steering wheel of the car.
3. Level 3 comprised of the car which is capable of performing some safety-critical functions in certain conditions and here we still require a human driver. It poses some risk between human and automated driving system.
4. Level 4 comprised of the car that can drive without having any inputs from the driver but needs some during the severe weather conditions.
5. Level 5 comprised of full automation in all the conditions. It also means the steering wheel which is optional in self driving car.
Challenges of autonomous vehicles comprised of many potential hazards which includes the inability of current infrastructure to respond to mistakes, errors which could lead to crashes, even cyber security threat of vehicles being hacked stealing all the confidential information, the unexpected malfunctions with the software systems which lead to accidents. Even challenges such as creating and maintaining maps for the self-driving cars seem to be difficult work. Here google’s self driving car works by relying on the combination of already prepared maps as well as the sensors that detect obstacles on the road....