Small-Scale Testbeds for Connected and Automated Vehicles in Research and Education

at IEEE ITSC 2024

Date: September 24, 2024

Room: Salon 19/20

Abstract

The growing importance of Connected and Automated Vehicles (CAVs) calls for innovative platforms to develop, test, and validate new algorithms and technologies. Small-scale testbeds offer an effective means to prototype and experiment with close to real-world conditions. Furthermore, these platforms tend to be powerful research demonstrators that motivate collaborations within and outside academia. However, these testbeds present unique challenges in terms of accessibility, reproducibility, and integration into educational frameworks. To better approach these challenges, this workshop aims to facilitate discussions on learnings and best practices. In particular, the workshop focuses on addressing these issues with an emphasis on leveraging small-scale testbeds for research and education.

Program

Time / Event
08:30 - 08:50	Arrival and Registration
08:50 - 09:00	Opening Remarks
09:00 - 09:30	A Research and Educational Scaled, Smart City Testbed for Real-Time Control of Autonomous Driving Systems Andreas Malikopoulos (Cornell University)
09:30 - 10:00	Cyber-Physical Mobility Lab: An Open-Source Testbed for Connected and Automated Vehicles Simon Schäfer (RWTH Aachen University) This talk will present our Cyber-Physical Mobility Lab (CPM Lab), an open-source, remotely accessible platform for Connected and Automated Vehicles (CAVs) focusing on multi-agent decision-making. The CPM Lab provides a deterministic computation model that allows for reproducible experiments. Using its digital twin, it can seamlessly extend its 20 small-scale vehicles (µCars) with virtual µCars. We created a research and teaching stack around the CPM Lab: undergraduate students study programming in the CPM Academy, graduate students study coordination and control of CAVs, and researchers compete in the CPM Olympics. The CPM Lab is one of the pioneering testbeds for CAVs and has been rebuilt at other universities, establishing its influence in academic settings.
10:00 - 10:30	Coffee Break
10:30 - 11:00	Autoware Reference Platforms: From Small Scale to Full Scale Autonomous Vehicles Simon Thompson (TIER IV)
11:00 - 11:30	SVEA and Recent Experiences With Small-Scale Vehicles in Industrial Collaborations Kaj Munhoz Arfvidsson (KTH Royal Institute of Technology) As the demand for Connected and Automated Vehicles (CAVs) continues to rise, scalable, cost-effective platforms like small-scale testbeds are becoming increasingly crucial for prototyping, testing, and refining CAV technologies. This talk will present the SVEA platform, an open-source, small-scale testbed designed for evaluating novel automation techniques and V2X use-cases. Drawing from real-world experiences, the presentation will highlight how the SVEA platform has been used in recent experiments with industrial partners. In addition, the platform has served as an educational platform for teaching automatic control. With these experiences, the talk will cover how to lead collaborative industrial projects, opportunities for innovation, and pathways for future industry-academic partnerships while using the platform in education.
11:30 - 12:15	Panel Discussion Moderated by Jonas Mårtensson

Organizers

Kaj Munhoz Arfvidsson

KTH Royal Institute of Technology

kajarf@kth.se

Jonas Mårtensson

KTH Royal Institute of Technology

jonas1@kth.se

Smart Mobility: AI and Intelligent Decision-Making for Future Transportation Systems

at IEEE ITSC 2024

Date: September 24, 2024

Room: Salon 19/20

Abstract

The workshop focuses on the transformative potential of artificial intelligence and advanced decision-making technologies in modern transportation. As urban environments evolve and demand smarter mobility solutions, this workshop will explore how AI, data analytics, and intelligent systems can enhance traffic management, road safety, and autonomous driving.

Through a series of expert-led discussions, the workshop will cover topics such as leveraging real-world vehicle data, the development of digital twins for transportation modeling, the integration of reinforcement learning with planning and control systems, and ensuring robustness and trust in AI explanations for autonomous vehicles. Participants will gain insights into the latest advancements in smart mobility, offering a forward-looking view on the future of transportation systems.

Program

Time / Event
13:30 - 13:35	Opening Remarks
13:35 - 14:30	IEEE NTDAS: Driving global roadways research with real-world vehicle data analytics platform David Goldstein and Lavanya Sayam (IEEE National Transportation Data & Analytics Solutions) IEEE's National Transportation Data and Analytics Solution is a powerful platform that provides a unique, robust, and high-quality transportation dataset combined with advanced analytics tools, enabling valuable insights to empower academic research and instruction. Equipped with 5 min granularity data for speed and travel time, for both trucks and passenger vehicles, the NTDAS supports a multitude of use cases across domains whether it be in logistics / supply chain, freight research, urban studies and planning, sustainability, equity, safety, environment, urban and rural dynamics, and many others. This session speaks to the impactful ways NTDAS can be incorporated into research and teaching, with specific use case examples relevant to the global audience.
14:30 - 15:00	Development a Digital Twin Visualizer for Intelligent Transportation Applications Yunchen Ge and Marcelo Contreras Cabrera (University of Alberta) A high-definition (HD) digital twin framework, which provides a real-time and realistic visualization of multi-modal perception required for state estimation, navigation, and decision-making for autonomous driving, will be discussed and presented in this talk. We will introduce a perception system utilizing synchronized sensing data, the process to generate frame transformations to construct accurate HD maps for navigation in dynamic environments, and a framework to readily customize essential map elements such as road features and static landmarks. Application of the visualizer for localization tasks will also be demonstrated.
15:00 - 15:30	Integrating Reinforcement learning, with planning, control, and human input Matthew E. Taylor (University of Alberta) Reinforcement learning (RL) is a popular method for solving complex decision-making tasks, particularly when the full problem is unknown at design time. However, RL agents learning from scratch may require much more data than is readily available. In this talk, we will discuss how planning, control methods, and human intervention can be leveraged to quickly learn high-quality control polices that can outperform other methods, and how such methods could be applied to problems in autonomous driving.
15:30 - 16:00	Coffee Break
16:00 - 16:30	Robustness Analysis of Interactive Explanations for Traffic Scene Understanding: Implications for Trust in Autonomous Driving Randy Goebel and Shahin Atakishiyev (University of Alberta)
16:30 - 17:00	Developing Scenes and Scenarios in Automated Driving Applications Harish Chintakunta (MathWorks) The development of scenes and scenarios plays a crucial role in the design and testing of automated driving applications. This presentation will explore the workflows used to create interoperable scenes and scenarios with commonly used driving simulation tools. We will cover designing scenes, building scenarios from recorded data, and simulating driving applications for early design and testing phases.
17:00 - 17:30	Panel Discussion Moderated by Ehsan Hashemi

Organizers

Mohammad H. Mamduhi

University of Alberta

vahid.mamduhi@gmail.com

Ehsan Hashemi

University of Alberta

ehashemi@ualberta.ca