It feels like autonomous vehicles have been a promise for about as long as flying cars. But, unlike flying cars, could the reality of self-driving cars be just around the corner?
Burkhard Goeschel, automotive industry thought leader and head of the FIA Commission for Alternative Powertrain technologies, seems to think so.
He notes that the key factors driving fully autonomous vehicles — or Level 5 autonomy — are improved safety, reduced fatalities and a more convenient and comfortable ride.
So, what’s holding back self-driving cars and how can simulation address the issue
The Biggest Roadblock to Autonomous Driving: Validating Functional Safety
“The complexity for Level 5 autonomy in city traffic with pedestrians and everything else is very difficult to handle,” says Goeschel.
“This requires uncompromising functional safety,” he adds. “There should not be one unaddressed issue — not one. A fatality should not happen, and functional safety is really key.”
A product is functionally safe when the designers can verify that it will operate in a safe manner regardless of the inputs and failures that may arise.
For an autonomous car to be functionally safe, its embedded software and electronics must be validated under billions of possible environmental conditions, such as variable:
- Intersection layouts.
- Traffic patterns.
- Pedestrian conditions.
- Cyclist conditions.
- Weather conditions.
The list is almost endless. It would take decades — or longer — to test these scenarios on the road.
Goeschel explains that this is where simulation will play an important role in the development of self-driving cars.
What Role Will Simulation Play in the Birth of Autonomous Vehicles?
“In autonomous driving, the vast number of use cases cannot be evaluated with hardware. It would require building hundreds of cars and driving millions of kilometers to see what is happening,” says Goeschel.
In the video below, Goeschel adds that simulation will help engineers reduce the number of prototypes needed to ensure the functional safety of autonomous vehicles.
With the help of simulation, engineers can test the car’s functional safety under thousands of scenarios in a fraction of the time needed to test these scenarios on the road.
“Embedded software is so important now and automotive complexity is so great that there is no other solution,” says Goeschel. “This can only be done using simulation.”
To learn of other roadblocks to autonomous vehicles that can be solved with simulation, read the full Goeschel interview: Connected and Autonomous Vehicles — A Thought Leader’s Perspective.
And to find out how ANSYS is helping autonomous vehicle pioneers win the race to market, read Engineering Safety for Autonomous Vehicles.