Print Friendly and PDF

From Advance Driver Assistant Systems (ADAS) to Autonomy

By Robert Hoffman

Team Leader | AI Specialist | Product Planner | Consultant

This post is part of the MICHauto Summit series, a collection of articles aimed to shed light on the evolving culture and careers in the automotive and mobility industries. This post is the view of the writer and does not reflect the views of MICHauto or the Detroit Regional Chamber. Learn more and register for the Summit today.

It was 1999, no one in the automotive industry was speaking the words radar, ultra-sonic, or high-definition mapping (at least on a regular basis). The word fusion was typically associated with the likes of a nuclear reactor, not the combining of data from multiple sensor sources. Little did the automotive industry know that this small camera company based out of Jerusalem would become the world’s most dominate force in automotive vision systems.

The rise of the mono camera would bring life to the core foundation of an autonomous vehicle, its perception. With refined algorithms such as lane detection, vehicle detection, and speed limit sign recognition the vehicle was now able to see the objects surrounding it. There was one major problem, it couldn’t accurately tell how far these objects were. The demand for accurate ranging data would lead to the implementation of sensors such as radar and LiDAR. This led to more advanced features such as adaptive cruise control, autonomous emergency braking, and lane keeping assist. As these systems started adding more sensors the complexity went up as well.

Meanwhile, the government had announced the DARPA Grand Challenge, a self-driving competition open to those willing to face the challenge with a reward of $2 million. The first competition (March 13th 2004) was held in the Mojave Desert and consisted of 150km that would prove to be very difficult, no one finished the race that year. The second race (October 8th 2005) a little larger at 212km, this time all the racers except for one completed the self-driving course. Stanford University ended up coming in first place and Carnegie Melon University taking second. It was these races that pushed the boundaries and drove the automotive industry to its current state today. Sebastian Thrun, leader of the Stanford Team, end up founding Google X and create a self-driving car (Waymo). Anthony Levandowski, who entered the only two wheeled motorcycle in the competition, would end up leading UBER’s self-driving unit.

By 2010 Mobileye had taken a majority market share in the mono-camera industry, their main competitors were Autoliv, Bosch, and Continental. There were additional companies such as Magna or TRW but they were all selling Mobileye systems. When Google took its self-driving car public it jump started the industry and started raising eyebrows on what the future of automotive would look like. Traditional Tier 1’s had research teams dabbling in this technology but not at this scale. The automotive industry was caught off guard, and what we saw next was anything less than incredible.

2017 MICHauto Summit

It wasn’t long after Googles self-driving car that Tesla announced its self-driving technology. Autopilot proved to be ahead of its time, society reacted different. Engineers were questioning the quality and reliability of the system. The industry knew Google’s self-driving car was a prototype project but this was going into production! When autopilot launched it was a huge success, media praised Elon Musk for creating the world’s first production semi-autonomous system.

The increase of sensors and electronic control units (ECU’s) led the industry to rethink its approach, the complexity of these systems had become overwhelming. It was not uncommon for OEM’s to have at least four suppliers on an active safety system. A more centralized solution to achieving full autonomy would be required, rather than having multiple microprocessors throughout the system it would be achieved by one centralized ECU.  This would raise a new set of problems, architecting software to achieve full optimization of the CPU/GPU. Increased complexity for systems engineering, determining when certain features are enabled and disabled. Managing complex data such as high-definition maps. Verifying the proper redundancies are in place to achieve functional safety. The list goes on.

The challenges the automotive industry faces will continue until autonomous vehicles are wide spread. With mainstream adoption of autonomous vehicles will come a new set of problems. What will society do with all the additional time available to them?

Urban areas will have a focus on ride sharing, eliminating the driver will reduce the cost of taking a vehicle to a destination. The interior of the vehicle will completely change, an increased number of displays, rotating chairs, tables, dimming windows, and augmented reality. Ridesharing companies will introduce vehicles that are specific for what customers are trying to achieve. This could be a productivity, work out, or massage car.

How marketing is conveyed to the user will be innovative, expect there to be preferred driving routes where businesses pay ridesharing companies, in return their vehicles pass their store. An augmented reality ad pops up on a car window when passing the local cheeseburger joint. Tap on that ad and get a 20% discount! If the customer doesn’t want to see these ads, no problem! Take the upgraded premium ridesharing vehicle.

Autonomous vehicles still face many hurdles, optimization of camera algorithms, implementation of artificial intelligence, and creating high-definition maps. Ridesharing companies will be the first to introduce autonomous vehicles available to the public, the data from these vehicles will be used to enhance the technology. The launch of the first commercially available autonomous vehicles for purchase won’t likely be until 2025. Once the public can purchase an autonomous vehicle it becomes more about what they are going to do with this additional time.

Robert Hoffman has extensive experience working for automakers and tier 1 suppliers, leading automated driving applications solutions. Robert is a blogger and consultant to the autonomous vehicle industry.”

More from the MICHauto Summit

OEMs of the Future Need More Talent From Silicon Valley