Can We Test the Safety of Autonomous Vehicles? Part Two

Please Share

Can We Test the Safety of Autonomous Vehicles?

Part Two

In Part One, we looked at Advanced Driver Assistance Systems (ADAS) and where autonomous driving fits within this framework. Let’s now take a look at what driving situations must “pass the test” and what technologies must work in perfect sync to deliver consumer confidence.

What needs to be tested?

The following ADAS Level 5 features are just a few examples of what needs testing, according to a US Congressional Research Service report on issues in testing and deploying autonomous vehicles (April 2021). They necessitate a diverse set of sensors, many of which are geared toward collisionavoidance:

  • Detecting pedestrians
  • Detecting objects and classifying them according to their physical attributes
  • Predicting the path of a moving object
  • Parking
  • Eliminating blind spots (360-degree viewing)
  • Monitoring road conditions in all types of weather
  • Monitoring accidents, congestion, and rerouting

What technologies are required to implement ADAS Level 5 features?

Keysight Technologies has done all the work for us with its excellent White Paper on testing AV’s. The following technologies all need to work together to put AV features through their paces:

  • Artificial Intelligence (AI)
  • Software
  • Many different types of sensors, including RADAR (Radar detection and ranging), LIDAR (Light detection and ranging), and cameras for vision
  • GPS systems for maps and navigation software
  • Ability to read road signs and traffic signals
  • Wireless communication systems, including 5G cellular as well as short-range bandwidths for vehicle-to-vehicle communications

And the list goes on.

Is autonomous driving a realistic possibility?

The jury is still out on whether self-driving cars will ever become a reality. The testing challenges are numerous and deserve their own article. Can simulations accurately represent reality? How many miles must be travelled before accepting a success rate of 100 percent? 99.999 percent success is not good enough. If a test fails, the conclusion is self-evident. If it passes, how do we know it will never fail? However, some automakers believe that they are on their way to realizing their vision of a fully automated vehicle. In the meantime, test equipment manufacturers will be kept on their toes.

Please Share

Leave a comment

Your email address will not be published. Required fields are marked *