Validation of scenario-based virtual safety testing using low-cost sensor-based instrumented vehicle

Abstract

Autonomous vehicle (AV) requires millions of miles on road to test the reliability of safety systems. It is also difficult to test the AV for critical scenarios which are rare but will endanger road users. Therefore, virtual safety testing simulation platforms are introduced to test the safety systems of the autonomous vehicles in critical scenarios. However, developing the virtual safety testing simulation platform requires information about the environment and driving data from the real world. Besides, it is challenging to build a system to collect driving data which is normally cost intensive especially in developing countries. Paradoxically, these developing countries have poor traffic environment which can provide valuable scenarios for safety testing test cases. Therefore, in this paper, a scenario-based testing using virtual simulation platform is developed using data captured by a low-cost sensor-based instrumented vehicle. The instrumented vehicle is built by low-cost off-the-shelf components for the testing purpose. The instrumented vehicle is used for validation process in IPG CarMaker’s vehicle model using SAE standards. Then, the validated vehicle model is used as an autonomous vehicle in IPG CarMaker for the virtual scenario-based safety testing. The whole validation process from data collection to data logging is carried out using various economic sensors instead of a single industrial system. This approach greatly reduce the cost of the instrumented vehicle and the result of the scenario-based testing shows that the virtual scenarios developed in IPG CarMaker can be used for validation purpose with actual scenarios using low-cost sensor based instrumented vehicle as low as 4% root mean square percentage error.