Semantron 22 Summer 2022

Autonomous vehicles

Level 4 autonomy has already been achie ved with Waymo’s self -driving taxi. Extensive progress in the autonomous industry has already been accomplished, and while the jump from level 4 for level 5 is significant, companies are building off previous iterations and have been carrying out specific machine learning exercises since the race for level 5 autonomy began. The same can be said for mapping; companies take a vehicle equipped with cameras and lasers and drive through different areas. These lasers send out pulses of light in order to pick up the landscape. Sensors read the reflected pulses to construct a 3-dimensional map of the world, and this information is then processed and can be inputted to future autonomous fleet. In this sense on board maps are much more matured than any other feature on a self-driving vehicle. In 2019 , ‘77 autonomous vehicles from thirteen China -based companies steered their way across 1.04 million kilometres (646,226 miles) of busy urban streets, which is a 153,600 kilometre (95,442.6 mile) increase from 2018’ ( Researchers make breakthroughs 2020). Elon Musk famously stated that autonomous cars will only have to rely on cameras in the future, but the performance of these cameras is dependent on the surrounding light levels and so at night, the reliability would be much less than during the day ( Elon Musk says full self-driving Tesla tech 'very close', BBC 2020 ) . In America, visible light produced from a vehicle’s headlamps is regulated by the National Highway Traffic Safety Administration (NHTSA) (Thakur 2017). As shown in the figure below, visible light is legally limited to roughly 60 metres in front of the vehicle, while the light barely illuminates outside the width of the road. This forces the cameras to work with infrared light (which is not regulated).

Infrared

Sensors

for

Autonomous

Vehicles.

Thakur 2017

For infrared cameras to work reliably in the dark, more infrared photons would have to be emitted. However, the safety of the human retina

comes into play due to the reflective properties of different materials, i.e., some materials will reflect the infrared photons at different intensities. Extensive research and testing will have to be conducted to prove that the levels of infrared photons used by automotive manufacturers are safe in variable conditions and environments. This process would be incredibly time-consuming as the tests will have to be conducted at different light levels, in different environments (such as a city centre and the countryside), as well as different nearby materials and surfaces. Another problem engineers will have to tackle is the onboard computer, specifically, its energy consumption. The onboard computer is the brain of the vehicle, and this system takes a large amount of power to operate at the required performance levels. Prototypes use around 2,500 Watts of energy. Although this might not be a problem initially, since self-driving cars/shuttles will be bound to cities, this energy-hungry system becomes problematic when travelling to destinations with minimal recharge stations. Wilko Stark, Mercedes- Benz’s vice president of strategy, said that ‘ To put such a system into a combustion- engine car doesn’t make any sense, because the fuel consumption will go up

22

Made with FlippingBook interactive PDF creator