Search Results

Testing vision-based advanced driver assistance systems (ADAS) in a Camera-in-the-Loop (CiL) bench setup, where external visual inputs are used to stimulate the system, provides an opportunity to experiment with a wide variety of test scenarios, different types of vehicle actors, vulnerable road users, and weather conditions that may be difficult to replicate in the real world. In addition, once the CiL bench is setup and operating, experiments can be performed in less time when compared to track testing alternatives. In order to better quantify normal operating zones, track testing results were used to identify behavior corridors via a statistical methodology. After determining normal operational variability via track testing of baseline stationary surrogate vehicle and pedestrian scenarios, these operating zones were applied to screen-based testing in a CiL test setup to determine particularly challenging scenarios which might benefit from replication in a track testing environment.

In order to validate the operation of advanced driver assistance systems (ADAS), tests must be performed that assess the performance of the system in response to different scenarios. Some of these systems are designed for crash-imminent situations, and safely testing them requires large stretches of controlled pavement, expensive surrogate targets, and a fully functional vehicle. As a possible more-manageable alternative to testing the full vehicle in these situations, this study sought to explore whether these systems could be isolated, and tests could be performed on a bench via a hardware-in-the-loop methodology. For camera systems, these benches are called Camera-in-the-Loop (CiL) systems and involve presenting visual stimuli to the device via an external input.

Evaluating the safety of Autonomous Vehicles (AV) is a challenging problem, especially in traffic conditions involving dynamic interactions. A thorough evaluation of the vehicle’s decisions at all possible critical scenarios is necessary for estimating and validating its safety. However, predicting the response of the vehicle to dynamic traffic conditions can be the first step in the complex problem of understanding vehicle’s behavior. This predicted response of the vehicle can be used in validating vehicle’s safety. In this paper, models based on Machine Learning were explored for predicting and classifying driver’s response. The Naturalistic Driving Study dataset (NDS), which is part of the Strategic Highway Research Program-2 (SHRP2) was used for training and validating these Machine Learning models.