We had a talk with Soma Muthumanickam, lead software architect for automated driving functions at ZF, about the perspectives of self-driving cars and their impact on the wellbeing of the consumers. Soma Muthumanickam reveals the biggest obstacles that are preventing widespread adoption of autonomous vehicles and shares his thoughts on how the transition to self-driving cars will work and who will win the self-driving race.
Why there is a need for self-driving cars?
When it comes to self-driving cars and their integration into society, there are two approaches. The first approach takes the human out of the loop. At this point, taxis become mostly as Robo-taxis. This approach will have a really interesting impact on the existing society. The other approach is a next-level system, which originally is a consumer product for selling in a particular target customer group.
The self-driving cars are having a great impact nowadays. However, there will be even more possibilities in the future. For example, Robo-taxis will be handier than normal cars. On average, people use their cars only 10% of the time during the day. The other 90%, their cars are staying in the garage or parking lot. Moreover, prices for parking are rather high that makes all the expenses even bigger.
This problem is also connected to the space-wasting. Even major cities have up to 20% of the space for all the cars on the streets. In addition, when people drive their cars alone, they also don’t use all the space of the car efficiently. These are the drawbacks of personal car ownership that can be changed with the introduction of the Robo-taxis.
Robo-taxis can drive 24 hours per day, they are easily accessible for people and also have a “trend factor”, which makes them really popular among customers. One of the benefits of the Robo-taxis is connectivity. These cars can be connected together in some type of hive mentality. So, the level of autonomy makes real change in the way people drive cars.
What technologies are crucial for autonomous driving?
When it comes to autonomous driving in general, we split the functional aspects into four components.
First if the perception component, which is basically sensing of the environment, creating its model. Here, the main functional technological base consists of cameras, sensors, radars, lidars, and ultrasonics, etc.
The second component if prediction, which is an anticipation of what are the different agents of the environment going to do. But this component is very tricky, as accurate predicting is not always so obvious as it seems to be. For example, it’s really difficult to predict if the pedestrian crosses the road, or stays on the edge of the road, or even changes his/her mind in the middle of the way. That is why there are many guessing alternatives involved to make the decision-making process easier for the Robo-car.
The third component is responsible for trajectory planning. This component consists not only of vehicle trajectory and the ways it can move safely but also of feedback from the different parties in the environment.
The last one is controlling the vehicle component, which is completely finished functional component.
But the functionality of autonomous driving is not limited to these components. They all are going into the operating system, then to the hardware, and then to the infrastructure. So, it is a complex process of combining the functions and collaborating on different levels of development, giving many opportunities to other industries to work with automotive companies.
What are the biggest obstacles that prevent the widespread adoption of autonomous vehicles? And how will the transition to self-driving cars works?
There are two different approaches to autonomous driving.
One approach is when we make self-driving cars as a service, which is basically a Robo-taxi. This works like Uber in some way, when you have an App, call for a cab and get an autonomous car that picks you up and takes you to the place you want. This is not ownership but a service perspective.
The other approach is ownership of the self-driving car. This means developing the strategies that cover the demand and can provide the quality supply. When a person buys a self-driving car, you as a company need to make all the calculations on how much this person is willing to spend on a self-driving car, depending on average wage and time of using it (for example, 10 000-15 000$ that you can charge for the whole autonomy package, otherwise people will not buy it). So, the pricing is one of the questions of this approach that need to be answered in the nearest time.
Both of these approaches have many obstacles to their adoption. When it comes to autonomy as a service, it is a completely new idea, which has to be fully autonomous to drive a person safely.
Moreover, technological improvement is also one of the main problems because we don’t have enough ideas and money for their introduction. All the technological initiatives are occurring all the time but their implementation is rather a long-term process.
The legal and ethical aspects, which need to be solved, are also fundamental questions for autonomous vehicles.
What OEMs and Tier-1s should do so that the driver entrusts his/her life to the self-driving car?
The really important aspect for them is that people can trust their car. They are also responsible for the visualization of the car’s perception, showing what the car is seeing in real-time, letting the driver and a passenger have the feeling that their car sees everything around. The other things will come over later. But, operating in specifically targeted places in the world, which are highly mapped, is a really difficult question for the next years of the development of this industry.
How can OEMs embrace technologies faster? In cooperation with startups – a solution?
It’s not the only way but it definitely helps to develop the technology. The automotive industry has always had special rules that could be applied only to the automotive industry itself. Because of the scale and the specifics of the operations, the automotive industry couldn’t always keep up with the startups in the IT sector.
However, now it is changing because the software is becoming more and more important for the consumer. Every driver tends to drive a car with a perfect software providence, without having a bad user experience at all. Now, all the OEMs try to ramp up their software hiring strategies, collaborating with the startups.
Could there be a solution to the Trolley Problem?
There is no universal solution to this problem as it is an ethical and legal aspect of all the countries in the world. My opinion is that there should be no company or entity that makes this decision but rather it should be decided on a more global scale. There are many versions of this Trolley Problem.
One of them is the most popular version of choice. Here you have to decide how to act if the car can heat an old lady or a young boy. There were many surveys done to solve this issue and the results were very interesting. For example, many Western countries prefer to save a baby and many Eastern countries prefer to save an old lady. There was a strong difference in opinions depending on the culture and mentality.
The other version is about saving the passenger or the pedestrian. Of course, you will try to do both, but what if you need to choose? This problem was also tried to be solved in different countries. For example, in China, people tend to save the passenger more than pedestrian at all costs, while in Japan, people tend to save a pedestrian at all costs. It means that these countries are not even geographically apart but have strongly different opinions.
This dilemma points to the important innovative strategies in countrywide ethics for autonomous driving and artificial intelligence. The industry will probably not only try to answer questions at a minor detail level but also change the ethics perception at a major detail level.
Who will win the self-driving race? And when we will see the self-driving cars on the streets?
I think, there will not be one winner because the field and the complexity involved in it are so huge that it’s unlikely to have one single winner. But I see what is happening now, when it comes to autonomy as a service, when you have some companies that are good in certain regions, having more big data self driving cars from those regions. Nevertheless, I’m sure that at the end of launching, it will be one winner – a consumer.
To conclude, there are two answers to the question of terms of launching. The first one is when you are talking about autonomy as a service, then it will happen in two-three years. However, it will not be a global but geographically isolated launch. For example, some places in the USA and some places in Europe will have autonomy as a service. But, when you are talking about buying a self-driving car, then we are a bit far away because of the pricing and scalable tech. So, it will happen in eight-ten years.
The adoption of autonomous vehicles can change the world of driving very soon. But what challenges are waiting for consumers and companies? The demand for autonomous driving not only shows the problem areas but also allows the rapid introduction of innovations that will be a part of autonomy all over the world. So, If we want to be sure that the consumer will definitely win in the end, we need to wait until the self-driving cars will be not only in use but also on the market.