Browse Topic: Automated vehicles
ABSTRACT Over time, the National Institute of Standards and Technology (NIST) has refined the 4Dimension / Real-time Control System (4D/RCS) architecture for use in Unmanned Ground Vehicles (UGVs). This architecture, when applied to a fully autonomous vehicle designed for missions in urban environments, can greatly assist in the process of saving time and lives by creating a more intelligent vehicle that acts in a safer and more efficient manner. Southwest Research Institute (SwRI®) has undertaken the Southwest Safe Transport Initiative (SSTI) aimed at investigating the development and commercialization of vehicle autonomy as well as vehicle-based telemetry systems to improve active safety systems and autonomy. This paper will discuss the implementation of the 4D/RCS architecture to the SSTI autonomous vehicle, a 2006 Ford Explorer.
New forms of highly automated Advanced Air Mobility (AAM) aircraft, such as electric vertical take-off and landing (eVTOL) vehicles, could transform transportation, cargo delivery, and a variety of public services. The National Aeronautics and Space Administration (NASA) conducted a series of flight demonstrations in collaboration with the Defense Advanced Research Projects Agency (DARPA) and Sikorsky Aircraft (a Lockheed Martin company) to progressively evaluate autonomous technologies. The autoland flight test research is a first in series for investigating the world’s first procedural descending-decelerating automated landing with vertical guidance Instrument Flight Procedures (IFP). The Sikorsky Optionally Piloted Vehicle (OPV) experimental UH-60 Black Hawk was used to evaluate a flight path’s four-dimensional trajectory (4DT) management into primitive commands and then follow those commands to a Point-in-Space (PinS) landing to the ground. All flight procedures were manually flown to the ground at 12 degrees with a 20-knot tail wind to ensure flight safety before automation was engaged. New and novel high precision approach procedures could pave the way for all future VTOL operations.
In this work, we present a lightweight pipeline for robust behavioral cloning of a human driver using end-to-end imitation learning. The proposed pipeline was employed to train and deploy three distinct driving behavior models onto a simulated vehicle. The training phase comprised of data collection, balancing, augmentation, preprocessing, and training a neural network, following which the trained model was deployed onto the ego vehicle to predict steering commands based on the feed from an onboard camera. A novel coupled control law was formulated to generate longitudinal control commands on the go based on the predicted steering angle and other parameters such as the actual speed of the ego vehicle and the prescribed constraints for speed and steering. We analyzed the computational efficiency of the pipeline and evaluated the robustness of the trained models through exhaustive experimentation during the deployment phase. We also compared our approach against state-of-the-art implementation in order to comment on its validity.
Letter from the Guest Editors
Items per page:
50
1 – 50 of 375