The simple act of driving can be an unproductive, dangerous, and time consuming activity, one that can be solved through the installation of autonomous technology within vehicles. This technology is considered to be among the most crucial breakthroughs in human travel that is being developed today; it is believed to have the capacity to create an improved and efficient driving experience by limiting fuel consumption, decreasing traffic congestion, and reducing wasted time during road trips.

One of the driving forces behind the creation of autonomous vehicles is safety. Autonomous technology promises safer travel compared to human-operated vehicles, as the cars are equipped with laser and video detection systems to control the car's speed and steering mechanisms while avoiding obstacles in the roadway. This blend of autonomous technologies promises to make driving 99% safer while also allowing the travelers to focus on other activities.1

These cars must detect and make rapid decisions to avoid objects in the roadway; the simple act of crossing an intersection requires the robotic cars to account for the inertias, right-of-way, and velocity of approaching vehicles.2 A major problem facing autonomous vehicles is the idea of real-time communication. As humans correspond face-to-face, these autonomous cars need to interact in real-time, allowing the cars to work together safely. However, this type of communication is unpredictable and extremely hard to maintain.3 Autonomous technology presents near endless benefits to automobile commuters; however, this technology faces not only current mechanical and software problems but also major legal and social issues. This technology needs to be perfected in every way possible before being released into city streets. Through my review of the autonomous technology within these computer-driven cars, I will explore the type of technology that operates these cars, how it operates the vehicle, the benefits created from this technology, and any possible legal and social concerns that arise from their use.

Developing Technologies: Seeing, Thinking, Steering

The ability for autonomous cars to see and judge risks in the roadway is vital to safe operation of the vehicle. An outstanding prototype of autonomous technology was created in 2007 by the Stanford Racing Team. Their robotic car Stanley, which won the DARPA Grand Challenge, operated solely on a software system that processed and converted visual data into appropriate driving commands.4 This software system uses an onboard sensors including lasers, cameras, and radar instruments to gather outside information from the road, allowing the robotic vehicle to observe and judge the approaching roadway;4 these sensors are placed on top of the vehicle. The combination of lasers and cameras allows for increased detection of obstacles by allowing both short and long range detection, respectively.4 As the cameras receive the long range images, the lasers allow the vehicle to detect the dimension of approaching objects that could harm the vehicle. Detection of hazardous obstacles is one of the easier aspects of autonomous driving; split second decision-making based on the detection system is harder to accomplish. An autonomous vehicle must use the information from the detection systems to determine if the road surface is safe for driving. Measuring the dimensions of detected objects allows the car to determine if they are true obstacles, such as roadway debris, or non-obstacles, such as grass and gravel. The researchers who helped build Stanley stated that the robot had trouble determining the difference between tall grass and rocks, which poses obvious difficulties in application.4 In addition to obstacle recognition software, autonomous vehicles require extensive algorithms to accomplish and maintain velocity, steering, acceleration, and braking—functions all controlled by the same system of detection and decision making.

Dynamically Guided Routes

Route guidance is core to autonomous vehicle technology, which is not safe and effective without a computed path. The purpose of route guidance is to gather information from outside sources (e.g. other vehicles, fleet signals) and stored data to create the most efficient route. However, this technology is hindered by the limited amount of information that can be stored within the vehicle due to static map conditions.5 Static conditions are defined as the basic components of individual roadways, such as the length of the road, speed limit, and pre-existing intersection signals. Using static systems can result in unreliable and slower routes due to an inability to account for dynamic road situations; for example, these static routes can be highly ineffective once an accident occurs on the roadways.

Generating accurate routes while on the road is another computationally challenging problem for autonomous technology.5 Due to the mobile condition of autonomous vehicles, current onboard computational power cannot compute and translate both long algorithms and dynamic conditions at the same time. Researchers attempting to create an algorithm must balance quick execution and efficient route creation with low computational power.

An additional problem arises from dynamic roadways. Dynamic roads are defined as streets that are always changing due to traffic jams, accidents, and construction.5 In his article on route guidance, Yanyan Chen stated that a good route is one that, although possibly not the fastest, is both reliable and acceptable to the driver’s needs. As a solution, Chen and his team created the Risk-Averse A' Algorithm (Figure 1). This algorithm suggests a risk-averse strategy that pre-computes factors that affect traffic (such as weather and time of day), accounts for dynamic traffic flow and accidents, and computes a low-risk and reliable route. The Risk-Averse A' Algorithm is widely accepted in the field of autonomous research as the most efficient form of computing reliable and adaptive directions. In fact, Stanley used this algorithm in the DARPA Challenge.4

The task of navigating an autonomous car through an intersection is not simple. The vehicles must be able to use algorithms to derive not only the distance from the car to the intersection but also its current inertia. Simultaneously, this information must be constantly compared with that of other vehicles. The two main challenges in crossing an intersection are establishing reliable communication with other vehicles as well as the dynamic, convoluted environment of intersections. For autonomous navigation to be possible, vehicles must communicate with each other to determine which car has right of way. When approaching an intersection, each car should propagate signals to the other vehicles, a failsafe in case oncoming cars are not detected by the visual and laser system (Figure 2). In theory, autonomous vehicles will discharge signals containing position and velocity information. At an intersection, approaching cars can detect and process this information to determine the appropriate mechanical move.

The dynamic environment of an intersection creates a whole new series of problems with the introduction of unknown variables. An autonomous system must be able to adapt, sense, and make decisions in short periods of time. The proposed ideas on how to navigate intersections use a decentralized navigation function, a method that has no need for long-range communication between vehicles. It enables cars to navigate independently while maintaining network connectivity and an overall goal. This function allows the car to account for dynamic traffic and improves the use of algorithms.2

Robotic Communication

The problem of real-time coordination between vehicles is a major obstacle that must be overcome for this technology to function safely on city streets and highways. Without reliable and fast communication, autonomous vehicles cannot navigate intersections, conserve energy, drive in safe formations, or create efficient routes. However, communication through wireless networks is not always reliable. Dr. Mélanie Bouroche from Trinity College, Dublin, stated that a “vehicle intending to cross an un-signaled junction needs to communicate in an area wide-enough to ensure that other vehicles … will receive its messages.”3 Figure 3 illustrates how the cars should disperse signals to communicate with other vehicles.

In the article “Real-Time Coordination of Autonomous Vehicles,” Bouroche, Hughes, and Cahill found a solution to this communication issue by creating a space-elastic communication model. A coordination model for autonomous cars allowed autonomous vehicles to adapt their behavior depending on the state of communication, ensuring safety constraints were never violated.3


Autonomous technology should improve daily travel by decreasing fuel consumption, traffic congestion, and accidents. The construction of new highways and streets to accommodate this technology would modernize and improve the efficiency of cities. Daily life could be enhanced, as driving time could be spent more productively. Autonomous technology can greatly improve everyday vehicular travel—but only if it is correctly implemented into society. Many problems still remain in the realization of autonomous vehicles: detection systems must be improved to effectively identify and avoid obstacles, algorithms need to be refined to quickly compute dynamic routes, and communication between vehicles needs to be drastically improved in order to avoid accidents. The legal and societal issues must also be addressed: will all vehicular travel be converted to automated travel? If so, will all citizens be forced to use technology that controls their movement? If not, will separate highways and roads be built? Who will fund this new creation of streets and roads? Who will ultimately control and maintain such a system? Autonomous technology has the potential to vastly improve travel, but it can introduce system vulnerabilities and malfunction. Self-directed vehicles must be thoroughly researched and tested before the technology can be implemented on city, state, and national streets.  


  1. Hayes, B. Am. Sci. 2011, 99, 362-366.
  2. Fankhauser, B. et al. CIS 2011: IEEE 5th International Conference, Qingdao, China, Sept 17-19, 2011; pp.392-397.
  3. Bouroche, M. et al. IEEE Conference on Intelligent Transportation Systems, 2006, 1232-1239.
  4. Thrun, S. et al. J. Field Robot. 2006, 23, 661-692.
  5. Chen, Y. et al. J. Intell. Transport. S. 2010, 14, 188–196.
  6. Bergenhem, C. et al. Sartre, 2008, 1-12.
  7. Dahlkamp, H. et al. In Proceedings of the Robotics Science and Systems Conference, 2006, 1-7.
  8. Elliott, C. et al. In The Royal Academy of Engineering, 2009, 1-19.
  9. Douglas, G. W. Unmanned Systems, 1995, 13, 3.
  10. Laugier, C. et al. In Proceedings of the IFAC Symposium on Intelligent Autonomous Vehicles. 2001, 10-18.
  11. Wright, A. Comm. ACM 2011, 54, 16-18.