Autonomous driving is probably one of the most promising applications of Machine Learning in 21st century. Different from the traditional cars, driverless car uses sensors, such as radar, camera and Lidar, to perceive surrounding environment and transfer surroundings into data. Then sensor data are used by algorithm and control systems to avoid obstacles, to determine the optimal driving path, and to keep the car safe.
Driverless cars don’t need drivers, so it can eradicate human errors . It can optimize the driving path according to live traffic data, so it can reduce the traffic jam, improve the traffic flow and increase the fuel efficiency. It can enhance the mobility for children, elderly and disables. It will revolute entire automotive industry and all businesses around it. Since many software companies, such as Google, Apple and Uber, started to enter the automotive industry, this trend forces traditional car makers to think about how to compete with these new players in the near future.
Volvo Cars path to autonomous driving
Volvo Cars’ competitive advantage has been the “safety” for the long time. It realized that automation will be the key part for the vehicle safety and started to focus on developing ADAS systems (advanced driver assistant system) in early 2000. ADAS are systems to assist drivers in the driving process, to reduce the human errors and increase the safety. From 2006 to 2017, Volvo Cars has launched 21 ADAS systems, such as collision warning with brake support, adaptive cruise control and lane departure warning. These technologies has put Volvo Cars on autonomous driving level 2, partial automation, since 2006.
Figure 1: levels of autonomous driving 
In 2017, Volvo Cars and Autoliv, the world’s largest automotive safety supplier, together built a joint venture, Zenuity, to specially develop the autonomous driving technology. In the same year, Volvo Cars launched “drive me pilot” in Sweden, the large-scale trails of autonomous driving carried by real customers on the real world. It can help Volvo Cars to not only validate its autonomous driving technology, but also learn customer behaviors through real testing. 1st of November, 2018, Volvo Cars and Baidu, the leading Chinese internet search provider, announced to collaborate together to develop fully autonomous driving cars, in order to mass produce them in China. 
Volvo Cars’ vision for 2020 is zero death. In order to achieve that, Volvo Cars is continuously focusing on developing ADAS and autonomous driving technology. Its plan for short term is to skip level 3, directly develop level 4 autonomous driving SUV in 2021.  For the medium term, Volvo Cars needs to optimize its operation and manufacturing process for autonomous driving cars, to work with government for the regulations and to change customers’ perception about the driverless cars’ safety.
There are some risks in the machine learning too, which Volvo Cars needs to consider. First, it is hard to know where the algorithm goes wrong when accident happens. For example, if a driverless car hits a person, we want to know why that driverless car makes the decision that hits that person. However, that decision algorithm is not decided by people, but learned from enormous data by machine itself. We don’t even know if it is the algorithm’s problem or that pedestrian’s responsibility. Second, machine learning algorithm designer’s bias might influence the machine learning’s result. For example, if the designer is racist, that designer might subconsciously put that bias into algorithm or feed biased data to train the machine learning. This would be very risky when driverless car with such biased algorithm is launched into the market. Therefore, in order to design a real safe driverless car, Volvo Cars has to have a very solid control system to detect those bias or problems inside the machine learning algorithms.
All auto OEMs and software companies are developing their own autonomous driving technology separately. However, in order to make it work, cars have to be able to communicate with each other and it means they need to use same communication protocol. Right now, different brands use different communication technologies and no one stands out to unify this protocol. How to address this problem? Should someone stands out right now to solve this problem, or wait until every car makers’ driverless car running on the street?
When car becomes autonomous, the relevance of the car ownership drops materially. BCG, a consultancy firm, states that by 2030 a quarter of passenger miles travelled on US will be a shared, self-driving electric vehicles, reducing the number of cars on city streets by 60%.  There will be less people buying the cars than right now. How this trend will influence Volvo Cars business model in the future?
Word Count 
- National Highway Traffic Safety Administration (NHTSA) Website, https://www.nhtsa.gov/technology-innovation/automated-vehicles-safety
- Volvo Cars Global News Room, “Volvo Cars and Baidu join forces to develop and manufacture autonomous driving cars”, 1st, Nov, 2018, https://www.media.volvocars.com/global/en-gb/media/pressreleases/240303/volvo-cars-and-baidu-join-forces-to-develop-and-manufacture-autonomous-driving-cars
- Volvo Cars Autonomous Driving website, https://www.volvocars.com/intl/buy/explore/intellisafe/autonomous-driving
- Economist, “Reinventing Wheels”, 3rd March 2018, https://www.economist.com/special-report/2018/03/01/autonomous-vehicles-are-just-around-the-corner