Simulation in Autonomous Driving

Knowledge Hub
Automotive Performance Engineering

Powered by

Simulation in Autonomous Driving


As automotive products are gradually becoming smarter, it is literally becoming impossible to test the complex systems loaded with electronics for undergoing physical testing. Companies are rapidly moving towards adopting virtual product development methods as new product introduction timeline is shrinking faster. There is a strong need for simulation methods to be used today as a standard across the automotive industry, as there is an increasing preference for safer roads. As the automotive industry moves towards deployment of autonomous driving, the next big challenge would be carrying out validation of various features emerging across domains. Autonomous feature of the vehicle is critical from a safety point of view, as numerous malfunctioning incidents have been reported during autonomous drives.

According to industry experts, over 90 % of simulation techniques are co-related with the actual test but need to be carefully deployed for achieving better results. Autonomous vehicles will include features and systems, including sensors & actuators that need to be built as per specifications of vehicle software, sensor response time and performance variation in varying conditions. Going forward, vehicles would be required to communicate with the user, other vehicles, and with infrastructure while dealing with electromagnetic obstructions. So, a better understanding of how the waves propagate across the system will be imperative.

Read about: Virtually validate and test advanced driver assistant systems during the early development phases

Thus, developing features as per real road conditions are crucial as virtual simulation methods can address regulatory challenges. Varying road conditions such as road types, curves, weather conditions, junctions are being explored by several organisations using 3D modelling, digital twins and also object modelling to stay as close as possible to real world scenarios. Over the years, vehicle dynamic models have matured. However, it is the sensor model that poses a key challenge. This model is still in the evolution stage and industry experts believe data-driven models could rope in maturity to sensor models. Due to the involvement of hardware, software and uncountable data, integration of artificial intelligence, machine learning and IoT can pave the way for efficient and robust solutions for conducting simulation activities. Autonomous driving is a complex ecosystem that will need validation of performance, safety and reliability tests, as a holistic solution through simulation can reduce road testing costs by up to 40 %.

Read More: Understanding the Move to Intended Functionality in Autonomy