A self-driving car was once an idea so futuristic and innovative it appeared a better fit within the pages of a science fiction novel than driving along our roads. However, with companies like Waymo currently expanding their test sites for so-called “robotaxis” to major cities including New York and Los Angeles, it is predicted this revolutionary service will be available to the public by 2024. In this module, I will produce a series of screen-based interfaces which enable interactions with autonomous vehicles guided by research.
(above) An example of a Self-Driving car in action
(above) An example of a Self-Driving car in action
An automated or self-driving car is a vehicle capable of sensing its environment and operating without human involvement. There are different degrees of driving autonomy ranging from “No Automation” to “Full Automation”. For example, the former requires a full-time human driver controlling all aspects of the dynamic driving task. This would include driving with both hands on the wheel and starting the car manually, as well as the responsibility of monitoring their environment despite the warning or intervention systems their vehicle may be capable of.
(above) Infographic highlighting the levels of Driving Automation
Gradually the scale begins to relinquish responsibility from the human driver as the car gains autonomy. For example, “Conditional Automation” allows the car to perform all driving tasks with the exception that the human driver can respond and intervene when necessary - e.g., performing an emergency stop when the vehicles system fails to observe its environmental conditions and act appropriately in a potentially dangerous situation. As previously noted, the latter: “Full Automation” allows the human driver to essentially become a passive passenger if they wish it. The fully automated car is capable of performing the entire driving task under all environmental and traffic conditions without its human passenger having to worry about stepping in and taking control of the vehicle.
Automated cars are equipped with technology including sensors, actuators, complex algorithms, machine learning systems, and powerful processors to successfully drive from one location to another safely. According to Waymo’s website, their cars are equipped with three different types of sensors that allow the car’s system to observe its environment and act accordingly.
(above) Diagram shows Waymo car sensor positioning
LiDAR (Light Detection and Ranging) sensors work by targeting a laser at any object or surface and measuring the time which it takes for the light to reflect back to the sensor. This paints a 3D picture for the car’s system which it can use to measure the size and distance of its surroundings, including curbs, street corners and other parked cars. As the site states, these sensors provide a full 360 degree scan of the vehicles surroundings within as far as a three hundred metre radius.
(above) Gif presents how the LiDAR views traffic
There are also cameras situated around the vehicle, described by the site as a “vision system”. The cameras provide the car’s system with a more detailed, sharper image of its surroundings to the quality of a CCTV camera. This allows the car’s system to identify for example, pedestrians and traffic signs containing information like speed limits and roadworks which the LiDAR sensors could not present. As seen below, this can provide the car system with data such as a pedestrian crossing in a short distance - allowing the car to adjust its speed and be prepared for the possibility of a civilian stepping into the cars path.
(above) Another example of the Waymo camera sensors acknowleding street signs from a great distance
Waymo cars are also equipped with a radar sensor which can observe and measure an objects speed and distance from the car. It enhances the data that the LiDAR and cameras have collected, ensuring that the system is prepared for sudden risks that may approach the vehicles body.
All the data collected by the sensors is fed back to the car software that is made up of different programming including artificial intelligence, machine learning and mapping like Global Positioning System (GPS). These systems analyse the data and provide it with meaning. For example, if the sensors detect a pedestrian approaching the cars vicinity at a specific speed and distance, the systems will use this information to decide a safe, appropriate speed and navigational path to adjust the car to. These systems then send instructions to the car’s actuators, which control acceleration, braking, and steering.
Hard-coded rules, obstacle avoidance algorithms, predictive modelling, and object recognition are also contained within the system to help the software follow traffic rules and navigate obstacles. Companies like Hyundai have also discussed how the car’s system may be able to communicate with technology like traffic lights and other vehicles on the road too.