A revolution is underway in the transportation industry. The rise of autonomous vehicles will transform the industry and society itself as much as the nineteenth century shift from horse-carriages to automobiles did.
However, developing autonomous vehicle technology is a formidable challenge. It requires ambitious new developments in sensing technologies, machine learning and artificial intelligence, that are not only unprecedented in the automotive industry, but in all other industries as well.
The chief problem in replacing human drivers with artificial intelligence is that of machine perception. An autonomous vehicle’s computer needs the ability to recognize other vehicles, pedestrians, road signs, road markings, trees, buildings, traffic lights and multitude of other things that we encounter everyday while driving, and that too in poor driving conditions such as in the darkness of the night, in rain and in snow.
The problem is nearly impossible to solve with traditional rule-based computer algorithms. Instead, neural-networks and machine-learning methods need to be used. In these methods the computer is trained rather than programmed. But driving is such a complex task that an immense amount of training will be needed to make a computer drive as safely and reliably as an average human. Analysis done by the Rand Corporation indicates that autonomous vehicles will need to be driven through billions of miles of road tests to train its artificial intelligence to the same level of safety and reliability as a human driver.
Toyota Motor Corporation’s president Mr. Akio Toyoda echoed the same need at the 2016 Paris Auto Show, projecting that 8.8 billion miles of testing, including simulation, would be needed for developing self-driving cars. In contrast Google recently completed a cumulative total of 2 million miles of testing with its fleet of self-driving cars over the past 6 years. At that rate, it will take millennia to develop viable self-driving cars.
Engineering simulation is a time-tested tool for accelerating technology development. Faced with the daunting challenge of billions of miles of road tests, ADAS and autonomous vehicle companies have quickly realized that simulation will be an essential technology for achieving development goals and timelines.
Simulation accelerates ADAS and autonomous vehicle systems development in six areas:
- Driving Scenario System Simulation — involving simulation of complete driving scenarios in a virtual world model with detailed physical behavior of sensors and vehicle dynamics
- Software and Algorithm Modeling and Development — involving development of ISO26262 qualified and AUTOSAR compliant control and HMI software with model based development tools
- Functional Safety Analysis — for ensuring safety of automated systems with reliability analysis methods that connect with simulation tools for verification
- Sensor Performance Simulation — involving accurate modeling of radars, V2X communication, GPS antennas, ultrasonic and other sensors with high-fidelity physics
- Electronics Hardware Simulation — for optimization of signal integrity, thermal-structural-electromagnetic reliability of electronics and mechanical hardware components of ADAS and autonomous systems
- Semiconductor Simulation — for optimization of power efficiency, power noise integrity and thermal-mechanical reliability of new ICs being developed for ADAS and autonomous vehicle applications
I have provided more details of each of these six areas in a new white paper entitled Fast-Tracking Advanced Driver Assistance Systems (ADAS) and Autonomous Vehicles Development with Simulation.