Autonomous vehicles (AV) will require an extensive system of advanced sensors, onboard computers, high-speed and high-bandwidth data networks, and wiring to connect everything. This complex network of cameras, radar, lidar sensors and electronic control units (ECU) will be responsible for detecting and interpreting dynamic environmental conditions to inform real-time driving decisions. This means gathering, processing and distributing gigabits of data every second to enable the algorithms and ECUs to respond to a rapidly changing driving environment.
The complexity and criticality of the electrical and electronic systems required for autonomous driving will dramatically increase the challenge of vehicle design and engineering. This is due to the extensive testing and validation needed to ensure the safety of these systems. Most estimates predict that autonomous vehicles will require billions of miles-worth of testing to ensure their safety. To remain competitive, manufacturers will need to incorporate the lessons learned through simulated and real-world testing into their autonomous vehicle designs.
The technological demands of fully autonomous vehicles are enormously challenging for engineers. Advanced sensor technology, high-speed and high-bandwidth data networks, and cutting-edge artificial intelligence are all crucial to the functional and commercial success of autonomous vehicles. The real challenge, however, begins when these advanced technologies are integrated into a single system that must perceive, communicate, and decide on a course of action.
A car with Level 2 autonomy, for example, may feature active cruise control, a lane departure warning system, lane-keep assist and parking assistance. This car also requires 17 sensors to enable its driver assistance systems. These sensors consist of ultrasonic, long-range radar, short-range radar, and surround cameras to monitor the vehicle’s environment. Furthermore, the computations performed by this car’s automated systems are relatively primitive. The lane-keep assist system, for instance, is only tasked with monitoring the vehicle’s position relative to the lines of the road. Should the driver begin to stray, the system will notify the driver or take corrective action, but ultimate responsibility for control of the vehicle lies with the driver.
A Level 5 autonomous vehicle will have complete responsibility for control over the driving task, requiring no human input. As a result, a Level 5 car will have more than 30 additional sensors of a much wider variety, to cover the huge number of tasks involved. On top of the ultrasonic, surround camera, and long- and short-range radar sensors of a Level 2 car, Level 5 will require long-range and stereo cameras, lidar, and dead reckoning sensors. The increase in sensors will increase the amount of wiring needed in the harness and the necessary computational resources to handle the gigabits of data being produced by the sensors.
During design, engineers will perform architecture and trade-off analyses to investigate architectural proposals, such as a centralized versus domain versus distributed architecture. For an autonomous vehicle platform, these analyses will need to account for hundreds of components and millions of signals while optimizing function locations, network latency, error rates, and more.
Despite these challenges, AV is a burgeoning market. At least 144 companies have announced AV programs, and annual spending on semiconductors for ADAS applications is projected to grow year-on-year. Some of these are major automotive manufacturers seeking to stay ahead of the coming industry disruption, but most are startups or companies from other industries seeking to enter a traditionally impenetrable market. These companies lack industry-specific experience and the engineering resources necessary to plow their way through the complexities of AV design. Even the major automotive OEMs will face problems that their legacy design flows are ill-equipped to handle.
This will be true especially as companies move their AV projects from research, development, and one-off prototyping, into full-scale production. Autonomous systems will need to be optimized for cost, weight, and power consumption, while adhering to the most stringent safety requirements the automotive industry has ever faced. To compete, these companies will need a new design methodology that enables young engineers to design accurate and optimized systems, which can only be done by capturing the experience and knowledge of veteran engineers. They will need generative design.
