It’s not wrong to assume that autonomous vehicles are futuristic technologies but what most of us don’t realize is that this concept debuted long ago; in fact a well dig into its history takes us to the world of Leonardo da Vinci and beyond. Since then, key breakthroughs covering land, sea and air, built a proper foundation for development of autonomous vehicles. And series of DARPA Challenges (2004 -2013) and Tesla’s Semi Autopilot success in 2016 kick started the race for autonomous vehicles around the world. Today, companies like Waymo, Ford, GM Cruise, Tesla, Mercedes, etc. are fiercely competing to build a safe and reliable autonomous vehicle.

The global autonomous vehicle market is estimated to reach USD 54.32 billionin 2019and is projected to reach USD 556.67billion by 2026, growing at a CAGR of 39.47%for the same period. But as the market seemed so prominent and as the world was moving from partial automation to full automation, two recent and major crashes of Boeing’s 737 Max 8 aircrafts shook the industries that rely on autonomous vehicles. Crashes of Lion Air and Ethiopian Airlines have left around 340 people dead in the last six months. 

Even though full details are not exactly clear yet, both crashes had one thing in common: the plane would not listen to the pilot’s command. And with several other non-fatal 737 Max incidents reported, there are concerns regarding how Boeing’s automated system functions. 

Boeing opted to mount engines of 737 Max 8 aircrafts further forward and higher than the previous Max series. But in some circumstances, the new design resulted in upward pitching of aircraft, so Boeing decided to integrate new hardware/software system called MCAS (Maneuvering Characteristics Augmentation System) to control upward pitching of the aircraft. 

In the case of Lion Air, faulty sensors reported that the airplane was stalling, even though it was not, which triggered MCAS to pitch aircraft down to gain speed, resulting in steep dive and a crash. The point worth noting here is that MCAS took total control of the aircraft and did not adhere to the pilot’s command to pitch up. In case of Ethiopian Airlines, similar events have been noted at similar time interval and its Angle of Attack data showed strong similarities with that of Lion Air. 

Both crashes were very much preventable, and involved only few simple steps to disable the MCAS and take manual control. But neither were the pilots trained, nor did Boeing provide sufficient resource and documentation to understand this simple yet complex system. These two incidents can be a good source of learning for other automated systems and vehicle manufacturers.

1. All the autonomous vehicles are dependent on Artificial Intelligence, while AI itself relies on sensors and software to function. A faulty sensor or software is enough to confuse the AI for a possible crash that is never supposed to occur. Thus, a major lesson is to have a mechanism ready that can identify whether sensors are working properly and which one is misreporting the autonomous environment.
2. Numerous sensors are essential for autonomous vehicles to function, so manufacturers should employ sensors that act redundantly in case of failures, and no single point of sensor failure should doom the entire vehicle. 
3. Boeing’s decision to alter engine-aircraft design to gain fuel efficiency as a competition to Airbus has received stern criticisms. Had the design not been altered, MCAS could have never been designed. Thus, Autonomous Vehicle Manufactures should analyze their new retrofits from various perspectives and implement them only if it is essential for autonomous capability and safety.
4. Even though AI and machine learning have developed rapidly in past few years, we have not reached a state where AI can take full control of our vehicles. The level of automation should not be completely relied to the point where human lack attention or speed of response. Human interruption and participation is still very much essential as a fraction of reaction time can be a difference between life and death. Autonomous vehicle pilots / drivers must be well aware about their systems and what their vehicles are expected to do. Through the incidents, it is well known that proper and adequate training and periodic reinforcement is very much essential.

There are much more lessons to be learned, but based on the recent crashes, cognizance of these factors is vital. If we look back, the reason our air travel has been so safe today is due to lesson learned from each and every air crash that took place in the past. We cannot afford similar scenario that each and every autonomous vehicle crash becomes a major learning point in the future. Self-driving vehicles and autonomous vehicles are our inevitable future and manufacturers should prioritize safety before profitability and unhealthy competition.

Grishma Prasad AdhikariSenior Research Analyst at A2Z Insights