Abstract— Autonomous cars (Self Driving Cars) is the new technology breakthrough in automotive history that uses detection technology and advanced control systems to navigate along roads and highways. Autonomous car include lidar, ultrasonic and radar sensors, video cameras, a computer processor to successfully drive among other cars on the road. Safer driving, increased mobility and reduction in labor costs are many advantages the production of self driving cars; however, although self driving cars reaps the benefits, many disadvantages come in play. The cons include the limitations of a computer processor that cannot process human-like thinking in congested environments, or the software security against hackers. Electrical and Computer Engineers are currently working on autonomous car systems to improve and make self-driving vehicles soon a reality. In addition, electrical and computer engineers are constantly implementing and testing the computer processing system to make it reliable, astute and effective.
I. Introduction
Autonomous cars have always been viewed as a thing of future; driving at very high speeds with the driver enjoying a full cup of coffee. With advances in technology, the future might arrive faster than we think. In 1925, the Houdina Radio Control created the first driverless car known as the Phantom Auto that was navigated by a driver in another car behind the autonomous car. This was accomplished through ultra wideband radio technology that sends out radio impulses resulting in movement of the vehicle.
In 1950, further advances were made when RCA labs created a “Smart Road” that uses sensors along the road to detect metallic vehicles. The road guided the car along an electrically charged cable where the coils attached to the front bumper picked up a signal and shifted the steering system. Although this was a successful trial test on specialized road, this innovation would cause major renovations on roads and highways across the country which would have been costly and ineffective
This topic is interesting because autonomous vehicles could revolutionize how we commute to places everyday. In addition, in the last hundred years, technology has soared to incredible heights that will soon do the unimaginable.
The discovery and progression of autonomous cars have several major advantages ranging from convenience to safety; however, many disadvantages come into play such as liability in fatal car accidents and imperfections in data processing. Challenges still remain for autonomous vehicles that include the ability to sense, process data and eventually react with an action.
II. Fundamentals
The design of an autonomous vehicle is both complex and integrated. In a published article by Pan Zhao, he states “the autonomous vehicle is a mobile robot integrating multi-sensor navigation and positioning, intelligent decision making and control technology.” In other words, to understand how an autonomous car works, we must first recognize the several components that work together to get the job done.
In 1981, Ernst Dickmanns, a professor at Bundeswehr University manipulated a Mercedes-Benz vehicle to drive by itself. Dickmanns used a standard industrial microprocessor that attains signals from the array of cameras that were installed and kept the car on track. In addition, an integrated transputer system was placed in the front and rear facing out of the window that assets situations such as a moving human, or an approaching obstacle. The cameras would gather information into the dynamic database, evaluate the situation, send a signal to vehicle control which then outputs an action. Moreover, the Mercedes Benz vehicle was equipped with a EMS (Expectation-based Multifocal Saccadic) vision system that used a vehicle eye, an array of cameras focused at different angles, for road detection which then sends a signal to gaze control system that “not only determine the viewing direction ad hoc for the present moment, but also plan and optimize the viewing behavior in advance for a certain period of time” wrote M. Pellkofer, a professor at the University of Bundeswehr Munchen.
Gaze control to an autonomous vehicle is as significant as a brain is to a human being; without the ability to saccade, quick eye movements between points, drivers are more prone to accidents. Engineers offered a solution and created the EMS vision which orients a camera onto a physical object, such as a driving car, which sends an algorithm into the Dynamic Database. If the car suddenly makes an unpredictable movement, for example, switching lanes or slowing down, the action will center the object on the camera image and send a signal to the system resulting in an action such as breaking.
The EMS Vision system is one of the most complex components of an autonomous vehicles that uses sensors and cameras to effectively navigate through traffic. An important component is the TaCC, a multi focal pan-tilt camera head, that implements multiple saccades on several heights and angles. TaCC is so powerful and essential that the device can identify and focus onto saccades on a range of distances within 160-390 milliseconds. Likewise, the MarVEye is a multifocal vehicle eye that has four different focal length cameras that could detect new objects and distance estimation between the camera and the foreign vehicle.
III. Examples
The EMS Vision system was pivotal in the beginning era models and is the basis of autonomous vehicle todays; however, engineers have created new and advanced systems that are ten times as powerful and effective then technology developed in the 1980s. Autonomous vehicles use machine vision algorithms that enables obstacle detection, a pivotal component to autonomous vehicles in driving. LIDAR, a constantly spinning sensor that sends out a laser beam to capture a full 360 degree view of the vehicle surroundings, gathers information into point clouds and generates a digital elevation map.
The LIDAR unit works with cameras that captures several images to determine the distance of objects, traffic light and signs. In result, these two components come together to solve problems using the Kuhn Munkres algorithm. Moreover, the Kuhn Munkres algorithm optimizes a combination of solutions to problems and offers a minimal cost solution. For example, a self-driving car is on a road with traffic lights, pedestrians and bicyclists. The LIDAR unit creates a constant image of the vehicles surroundings; meanwhile, the cameras are focusing on these possible hazards that may affect the path of the car. The Kuhn Munkres algorithm compares these two data points constantly and if a bicyclist makes a reckless decision of crossing in front of the vehicle, the camera reads this hazard as a threat and sends a different signal which changes the path of comparison within the algorithm known as cost matching. “The matching cost is computed as the arithmetic mean of normalized costs of pairs when more than two entities are involved. Since those are heterogeneous by nature, each source pair has its own features subset, classifier and logistic function” says Marco Allodi in his research about Machine Learning for Autonomous Vehicles.
IV. Promise and Limitations
Autonomous technology has been implemented into state of the art aircrafts and foreshadows the progression of autonomous vehicles in the near future. What was once considered exclusive is now being integrated into vehicles that humans use everyday. Autonomous vehicles have promising qualities, such as advancement of technology, traffic reduction, higher mobility, but are also subject to glitches and errors that could potentially risk people’s lives.
The number one cause of accidents is distracted driving meanwhile ninety percent of crashes are the result of human error. Since computers, sensors and cameras are working constantly during the use of autonomous vehicles, the chances of error is minimal. In result, the safety of driving will dramatically increease and could save thousands of lives. Moreover, engineers have computed integrative algorithms that use precise measurements to perfectly stop a vehicle from a obstacle or hazard. Companies such as Uber, Waymo and Tesla have been working on self driving vehicles and are competing against each other including cab services. Competition not only provokes economic growth but also creates new technology and services for the public community. Another promise is enhanced mobility for people who have a mental or physical disability, low income citizens or impaired driving. Autonomous vehicles will reduce traffic congestion and offer a smoother ride and increased cruising speed. Lastly, labor costs will considerably decrease forcing car companies to change their approach to production and distribution of vehicles.
However, autonomous vehicle technology has only recently been increasing at an exponential growth meaning there are many opportunities for error. One limitation factor is the idea of liability in a accident involving a autonomous vehicle. In 2016, a Tesla driver was using the car’s autopilot option before he was killed in an accident. Question of liability still remains today because the driver was given several automated warnings to keep hands on the wheel however Tesla’s safety features might have been capable to stop itself.
Another limitation is hardware reliability and computer security. This limitation is pointed out by Michal Braverman-Blumenstyk, “But tomorrow's autonomous cars will be far more vulnerable because they will be networked. Some of the functionality of connected cars can be accessed remotely—velocity adjustment” This produces a serious situation because hackers are becoming more creative to breach security. Hacking an autonomous vehicle could cause serious casualties to both the passenger and pedestrians that could potentially compromise the pursuit of perfecting the autonomous car.
V. Roles of Electrical and Computer Engineers
Electrical and Computer Engineers (EECEs) are essential to the development and implementation of autonomous vehicle’s technology. Engineers are given a real-life problem and must solve it by weighting the pros and cons of every solution. For example, electrical and computer engineers are focused on how to take the data from the sensors, process the data and finally perform an action. Test engineers are vital in this scenario because a self-driving car are bound to make mistakes and must be tested before they encounter real life situations and put human life at risk. Moreover, an electrical and computer engineer must face ethical and moral problems such as car data breaching and split-second driving decisions and provide a mathematical and science approach to fix these issues. This is where design and research engineers flourish and invent algorithms and codes to accomplish their task. Engineers work together in a team to output the best solution they have in each case.
A large task for engineers behind the vehicles system is programming. Engineers must also have very strong skills in power electronics that is highly integrated into today’s vehicles. “Energy-efficient power conversion is playing a vital role in power electronics” says Chowdhury, an IEEE member.
In addition to EECEs being pivotal in the development of autonomous vehicles, civil engineers have also proven to be an important aspect. In the next ten years, autonomous vehicles companies are projected to roll out their cars to the public. This brings civil engineers into the picture because they will need to solve how to reconfigure the road space and help the sensors have an advantage in safety sensing. The LIDAR unit and cameras used rely heavily on recognition of signs and posting therefore civil engineers must redevelop and produce new technology to promote the advance of autonomous vehicles. Lastly, civil engineers are also constructing ideas of highways solely for driverless cars which will impact driving behavior and mobility.
VI. Summary
Autonomous cars is a field that has been gradually prominent in the last twenty years and will prove to revolutionize mobility. A basic autonomous system consists of sensors and cameras that send data to a processor that is run through an algorithm which results in an action. Many challenges remain for EECEs and other engineers ranging from technical obstacles to ethical problems. The promises are advanced mobility and decrease in automotive accidents; however, some potential obstacles are system security and decision making in split-second decisions. Electrical and computer engineers are responsible for designing algorithms, codes and integrated circuits while facing restraints in design and obstacles.