Live Demonstrations

Texas SMARTTrack (Safety, Mobility, and Autonomy Research and Testing - TST) is a three-tiered testing facility. Tiers 1 and 2 will be located on the UT Austin Pickle Research Center campus. Tier 3 is an open test bed on public roadways in Austin. This world-class proving ground will bring transportation agencies, university researchers, and private industry together to improve traffic safety, operations and management through smart transportation infrastructure and automated vehicles. TST will provide a safe and controlled proving ground to test emerging technologies and serve as a regional certification center to develop standards. This demonstration allows participants to learn more about TST. The 3D simulator in today's demo lets you virtually experience driving Tier 1 of TST.

In this demo, a novel driver-in-the-loop instrumental system for automated vehicle research and validation is delineated. Such a setup synergizes a cutting-edge six-degrees-of-freedom moving-base vehicle simulator with an immersive video-audio system, an advanced dSPACE SCALEXIO real-time hardware-in-the-loop simulation computer that runs high-fidelity and customizable digital twins of vehicle systems, sensors, roadway, and traffic, a dSPACE MicroAutoBox, as well as a biosensor suite comprising electroencephalography instrument for brain activities monitoring, electromyography apparatus for forearm muscle responses measurements, and an eye-tracking device for gaze and head movements tracking. The capabilities and versatility of the novel instrumental setup are showcased by demonstrating its utilization in some research projects.

In this eye-opening demonstration, we shed light on the susceptibility of Deep Learning models to Adversarial Attacks. Our groundbreaking research introduces an innovative solution – a robust deep-learning model that harnesses the power of Hybrid Classical-Quantum techniques. Throughout this demonstration, we meticulously assess the real-time performance of classical models in contrast to our cutting-edge Hybrid models, specifically in traffic sign detection.

This demonstration highlights the hidden risks of hardware trojans within intelligent transportation systems. Our research underscores the essential requirement for robust security measures to protect public safety and critical infrastructure from potential disruptions.

In this demonstration, we unveil a cutting-edge solution for the real-time detection of sophisticated Global Navigation Satellite System (GNSS) spoofing attacks for autonomous vehicles. Our innovative technology leverages advanced sensor fusion and machine learning algorithms to identify GNSS spoofing attempts as they occur. By showcasing our system's ability to protect navigation systems, we underscore the vital importance of safeguarding GNSS-based navigation systems against evolving cyber threats in an increasingly connected world. Witness firsthand how our solution ensures the reliability of GNSS data, preserving the security of location-based services in real-time.

The demo will show a new on-road XiL platform which enables on-road testing of connected and automated vehicles (CAVs) by blending the in-situ real and virtual worlds with mixed reality. The CAVs can be tested on a real road to interact with an unlimited mixture of real and virtual objects and scenarios in a very efficient, cost-effective and safe way. It can be applied to the testing of both autonomous and human-involved CAVs.   

In this compelling demonstration, we unveil the seamless integration of DJI drones with Amazon AWS cloud services. Our primary objective is to showcase real-time data collection capabilities, where data collected by DJI drones is seamlessly transmitted and stored within the robust infrastructure of Amazon AWS. This demonstration highlights our mission to craft a comprehensive end-to-end solution designed to gather data from drones efficiently and intelligently save it across various Amazon services, all while adapting dynamically to the unique demands of sensor data.

A real-time vision-based approach to improve pedestrian safety through the accurate detection of pedestrians and the generation of PSMs. The vision-based PSMs are generated in real-time (every 100 milliseconds), and these generated PSMs are used to improve pedestrian safety by developing and sending safety alerts in real-time from a C-V2X device to connected vehicles within its communication range.

One critical challenge for the deployment of autonomous driving in the real world is their safety evaluation. Most existing driving systems are trained and evaluated on naturalistic scenarios collected from daily life. However, the large population of cars, in general, leads to an extremely low collision rate, indicating that safety-critical scenarios are rare in the collected real-world data. Safebench is an evaluation platform that contains thousands of safety-critical driving scenarios generated by machine learning algorithms, which evaluate autonomous vehicles under realistic and critical scenarios. In this interactive demo, we will demonstrate safety-critical scenarios and let the users control a car in SafeBench to avoid collision in safety-critical scenarios. 

A holistic risk framework is used to independently monitor the decision-making of AVs and track their risk score in real-time. This generates both a reliable record of its safety score, and an early indicator if its safety margins are insufficiently robust. Use of the safety monitor ensures that non-linear instabilities in safety are detected and controlled to within a safe state, which satisfies the regulatory requirements for field-monitoring and event monitoring. Demonstrations of the risk framework are presented using naturalistic traffic studies in various environments.