 |
3 books on Self-driving Cars [PDF]
July 25, 2025 | 22 |
These books explain technologies used in self-driving cars, including computer vision, sensor fusion, machine learning and vehicle control systems. They explore the challenges of real-world navigation, safety, regulatory compliance and human-vehicle interaction.
1. Autonomous Intelligent Vehicles
2025 by Dr. N. Rama Devi, Dr. V. Rathinam, Kiran Raj Goli, Dr. T. Hussain
This is a GPT-generated human-assisted text, which mainly talks not about Autonomous Intelligence, but about related technologies (like sensors, control systems, communication and safety protocols). But it's interesting to read what GPT thinks about the current state of self-driving technologies. In particular, I was interested in the Natural Language Understanding in AV section. Here, GPT limited itself to recognizing driver commands, such as "let's go home." But what if we use language understanding to train an autopilot? Today, autopilots create their inner world from visual and other sensory information. But human texts also contain a lot of information about how to drive a car, what dangers there are and how to react to them. Of course, the language modality will greatly complicate the autopilot's inner world and make it less fast and reliable. However, perhaps more versatile and smarter on new roads. In any case, such an experiment would be interesting for understanding the benefits of LLM for autonomous driving.
Download PDF
2. Self Driving Car: Solving Full Self-driving Need Solving Real-world Artificial Intelligence
2022 by Fouad Sabry
The author of this book says that he wrote it with the help of Google - he sat for a long time and rewrote all lines that Google search gives in automplete when you enter "self-driving car" into it. Thus, he collected thousands of user queries - what questions about autonomous cars they are looking for - and then collected the answers to these questions - and voila - the book is ready. Perhaps the only interesting thing in it is a screenshot of the state of affairs / achievements of the self-driving car market at the time of writing the book (2022). According to the author, cars operating at autonomy level 3 and above continue to make up an extremely small percentage of the market. Waymo became the first company to provide the public with driverless taxi rides in December 2020. In March 2021, Honda became the first manufacturer to introduce a legislatively approved Level 3 car. But so far, consumers in the US do not currently have access to cars that can drive themselves. "Full stop," says the author. And this is despite the fact that automated driving systems (ADS) have been researched and developed since the 1920s. Most modern cars are equipped with safety systems such as cruise control, lane departure warning and emergency braking. But these are only driver assistance technologies, as they still require human control. Fully automated cars should be able to drive themselves, without any driver intervention.
Download PDF
3. AI-enabled Technologies for Autonomous and Connected Vehicles
2022 by Yi Lu Murphey, Ilya Kolmanovsky, Paul Watta
This book digs deeper into the fundamentals of building an autopilot for a self-driving car. The interaction between the autopilot and the environment can be modeled as a stochastic Markov decision process (MDP) with a human driver as a teacher. Data can be collected from a set of experienced driver responses and used to conduct reinforcement learning to develop optimal driving strategies. The driver's reward function can be approximated using a deep neural network. Deep reinforcement learning (DRL) is the foundation of autonomous cars, serving as a decision-making method in a hierarchical control system. However, the main problem of decision making and motion planning for DRL-based AVs is to prevent unsafe actions. The essence of the problem lies in the probabilistic nature of reinforcement learning: maximizing the reward function does not always guarantee the safety and reproducibility of decisions. The book describes a method that consists of limiting the output of the reinforcement learning algorithm within a safe loop defined by a deterministic decision-making strategy. To improve the security of DRLs, human-created rules are used. On the one hand, this allows combining human experience with machine learning methods, but on the other hand, such rules may be too conservative and limited only to the scenarios envisaged by the developers.
Download PDF
How to download PDF:
1. Install Gooreader
2. Enter Book ID to the search box and press Enter
3. Click "Download Book" icon and select PDF*
* - note that for yellow books only preview pages are downloaded
1. Autonomous Intelligent Vehicles
2025 by Dr. N. Rama Devi, Dr. V. Rathinam, Kiran Raj Goli, Dr. T. Hussain
This is a GPT-generated human-assisted text, which mainly talks not about Autonomous Intelligence, but about related technologies (like sensors, control systems, communication and safety protocols). But it's interesting to read what GPT thinks about the current state of self-driving technologies. In particular, I was interested in the Natural Language Understanding in AV section. Here, GPT limited itself to recognizing driver commands, such as "let's go home." But what if we use language understanding to train an autopilot? Today, autopilots create their inner world from visual and other sensory information. But human texts also contain a lot of information about how to drive a car, what dangers there are and how to react to them. Of course, the language modality will greatly complicate the autopilot's inner world and make it less fast and reliable. However, perhaps more versatile and smarter on new roads. In any case, such an experiment would be interesting for understanding the benefits of LLM for autonomous driving.
Download PDF
2. Self Driving Car: Solving Full Self-driving Need Solving Real-world Artificial Intelligence
2022 by Fouad Sabry
The author of this book says that he wrote it with the help of Google - he sat for a long time and rewrote all lines that Google search gives in automplete when you enter "self-driving car" into it. Thus, he collected thousands of user queries - what questions about autonomous cars they are looking for - and then collected the answers to these questions - and voila - the book is ready. Perhaps the only interesting thing in it is a screenshot of the state of affairs / achievements of the self-driving car market at the time of writing the book (2022). According to the author, cars operating at autonomy level 3 and above continue to make up an extremely small percentage of the market. Waymo became the first company to provide the public with driverless taxi rides in December 2020. In March 2021, Honda became the first manufacturer to introduce a legislatively approved Level 3 car. But so far, consumers in the US do not currently have access to cars that can drive themselves. "Full stop," says the author. And this is despite the fact that automated driving systems (ADS) have been researched and developed since the 1920s. Most modern cars are equipped with safety systems such as cruise control, lane departure warning and emergency braking. But these are only driver assistance technologies, as they still require human control. Fully automated cars should be able to drive themselves, without any driver intervention.
Download PDF
3. AI-enabled Technologies for Autonomous and Connected Vehicles
2022 by Yi Lu Murphey, Ilya Kolmanovsky, Paul Watta
This book digs deeper into the fundamentals of building an autopilot for a self-driving car. The interaction between the autopilot and the environment can be modeled as a stochastic Markov decision process (MDP) with a human driver as a teacher. Data can be collected from a set of experienced driver responses and used to conduct reinforcement learning to develop optimal driving strategies. The driver's reward function can be approximated using a deep neural network. Deep reinforcement learning (DRL) is the foundation of autonomous cars, serving as a decision-making method in a hierarchical control system. However, the main problem of decision making and motion planning for DRL-based AVs is to prevent unsafe actions. The essence of the problem lies in the probabilistic nature of reinforcement learning: maximizing the reward function does not always guarantee the safety and reproducibility of decisions. The book describes a method that consists of limiting the output of the reinforcement learning algorithm within a safe loop defined by a deterministic decision-making strategy. To improve the security of DRLs, human-created rules are used. On the one hand, this allows combining human experience with machine learning methods, but on the other hand, such rules may be too conservative and limited only to the scenarios envisaged by the developers.
Download PDF
How to download PDF:
1. Install Gooreader
2. Enter Book ID to the search box and press Enter
3. Click "Download Book" icon and select PDF*
* - note that for yellow books only preview pages are downloaded
