Visible to the public 


All Mini-Workshops will run concurrently from 3:45 p.m. - 5:45 p.m. on Thursday, November 21st.
Locations will be announced at the meeting. Please check the signage outside the meeting room for the location.

1. Coupled Communications and Autonomy Challenges in Connected Autonomous Vehicles
2. Autonomous Racing: Machine Learning for Perception, Planning and Control
3. CPS for Agriculture
4. Learning for Control
5. Medical Cyber Physical Systems: Current Research and Future Visions
6. Research Challenges in Smart and Connected Communities

1.   Coupled Communications and Autonomy Challenges in Connected Autonomous Vehicles

Organizers: Raghvendra V. Cowlagi and Alexander Wyglinski (Worcester Polytechnic Institute)


Vehicle-to-vehicle (V2V) wireless connectivity is the next frontier in road transportation, which will greatly benefit the safety and reliability of autonomous cars. Basic safety messages passed over V2V connections can help cars anticipate the movements of other vehicles, and consequently lead to better collision avoidance technologies. In the future, connected autonomous vehicles (CAVs) equipped with both self-driving technology and V2V connectivity, will lead to vastly improved road safety and efficient traffic flow.

Despite its promise, research on CAV technology is currently fragmented across multiple communities, due to which cross-cutting topics such as the interdependence of planning/control and V2V connections are poorly understood. The intention of this workshop is to bring together a diverse group of speakers/panelists and audience members to shed light from different perspectives on CAV state-of-the-art and challenges. Researchers in different focus areas of CAV technology typically attend different technical conferences. Therefore, this PI Meeting is an ideal venue for such researchers to assemble, share perspectives, and forge collaborations for future inter-disciplinary research.

This workshop addresses several research questions related to CAV technology, including:

  • Fusion of data from onboard sensors and V2V connections.
  • Flexible frameworks for motion-planning and control using data from onboard sensors and V2V connections.
  • The use of software-defined radio for V2V connections, including spectrum sensing and beamforming technology.
  • Intent estimation of surrounding vehicles using onboard sensor data and V2V basic service messages.
  • Impacts on autonomy of network delays and/or dropped packets in V2V connections.
  • Identifying in real-time the most informative V2V connections among those in range of communications.
  • Data-driven improvements in efficiency of heterogeneous vehicle fleets.
  • Designing CAVs capable of large-scale coordination (mass platooning).
  • Fundamental challenges associated with using V2V communications as a “sensor” for autonomous operations.

Program Agenda:

3:45pm - 3:50pm: Welcome and Introduction
Profs. Raghvendra V. Cowlagi & Alexander Wyglinkski (Worcester Polytechnic Institute)

3:50pm - 4:05pm: "Cooperative perception and cognition for connected and autonomous vehicle platoons"
Prof. Yaser Fallah (University of Central Florida)

4:05pm - 4:10pm: Audience Q & A

4:10pm - 4:35pm: Presentation on the topic of improving efficiency in heterogenous electric vehicle fleets
Prof. Fei Miao (University of Connecticut) & Desheng Zhang (Rutgers University)

4:35pm - 4:40pm: Audience Q & A

4:40pm - 4:55pm: "Modeling Options for Networks with Delay"
Prof. Matthew Peet (Arizona State University)

4:55pm - 5:00pm: Audience Q & A

5:00pm - 5:25pm: "Selective Listening: Control for Connected Autonomous Vehicles in Data-Rich Environments"
Profs. Raghvendra V. Cowlagi & Alexander Wyglinski (Worcester Polytechnic Institute)

5:25pm - 5:45pm: Panel discussion featuring organizers and invited speakers

2.  Autonomous Racing: Machine Learning for Perception, Planning and Control (Formerly: CPS Community Research in Safe, Fair, and Agile Autonomy)

Organizers: Rahul Mangharam (University of Pennsylvania) and Houssam Abbas (Oregon State University)


The workshop will cover five course modules that build on each other and include three races for evaluation:

  1. Introduction to ROS, F1/10 & the Simulator: Introductionto self-driving hardware and full software stack, automatic emergency braking, LiDAR, rigid body transformations, Laplace domain dynamics, and PID control for wall following.
  2. Driving using Reactive Methods & RACE!: Build vehicle, tune electronic speed controller, and implement reactive driving methods such as 'follow the gap' and complete Race 1.
  3. AV Mapping & Localization: Understand key elements of SLAM by implementing scan matching and particle filters, use Google Cartographer SLAM, implement pure pursuit driving, and complete Race 2 using maps.
  4. AV Planning: Learn about Moral Decision Making for autonomous systems, implement planning with with rapidly exploring random trees (RRT), and understanding model-predictive control (MPC) for raceline optimization.
  5. Learning & Computer Vision: Design and implement algorithms for detection and pose estimation, reinforcement learning and visual feature extraction.
  6. F1/10 Grand Prix!: Race 3 will include a project to implement racing strategies which combine fast perception, agile planning, and aggressive control for race optimization.


The F1/10 Autonomous Racing is a growing community of over 36 US universities interested in machine learning engineering for perception, planning and control. This year, we are introducing an Open Course Kit which will enable you to teach this exciting course covering the foundations of autonomous systems. This workshop will go over the lectures, labs, code and support forum to help you get started with F1/10. To get started, read the course overview slides -

F1/10 has hosted five international racing competitions and we will introduce the competitions at CPSweek (April 2020) and IFAC World Congress (July 2020). More details at

3. CPS for Agriculture

Organizer: George Kantor (Carnegie Mellon University)


Cyber-Physical Systems are being applied agricultural research and production in numerous interesting ways, addressing a wide range of topics such as plant phenotyping, big data analytics, precision livestock management, and robotic automation. This session will feature short invited talks from current CPS PIs working on NIFA projects to provide a sample of the important challenges and the approaches used to address them. There will also be a moderated panel discussion, followed by plenty of unstructured conversation time to build community among researchers interested in working in the agriculture space.

Program Agenda:

3:45pm: Introduction
George Kantor (Carnegie Mellon University)

3:47pm: Opening Remarks
Steven Thomson (NIFA-CPS Program Director)

3:55pm: AI, Robotics, and CPSs for Agriculture
Manoj Karkee (Washington State University)

4:10pm: Leveraging Animal Nutrition to Enhance Sustainability
Robin White (Virginia Tech)

4:25pm: CPS-enabled Soil Information System for Digital Agriculture
Nuwan Wijewardane (University of Nebraska-Lincoln)

4:40pm: Large-Scale Monitoring, Expert Input, and the Ecological Impact on Autonomy in Agriculture
Ryan Williams

4:55-5:25pm: Panel Discussion (all speakers, moderated by George)

5:25pm-5:45pm: Open Discussion and Networking

4.  Learning for Control

Organizer: Michael Zavlanos (Duke University) and George Pappas (University of Pennsylvania)


In current Cyber-Physical-Systems (CPS) supervised learning or unsupervised learning are typically used to analyze data. Nevertheless, in many such systems rules cannot be determined in advance and these data mining techniques are not directly applicable due to the dynamic nature of the data, their large volume that prohibits labeling in practice, and the fact that these data are added to the system piece by piece and not altogether in advance. On the other hand, control of CPS is usually done in a model-based manner, where a desired control policy is computed from a high fidelity system model that has been derived at design-time, and potentially may be updated at runtime. However, this approach is not suitable for highly dynamical CPS, that potentially represent systems of systems whose spatial and temporal configurations may rapidly change. Consequently, it is critical to facilitate design of data-based controllers, with strong performance guarantees, in a way that allows for natural runtime control adaptation.

The purpose of this workshop is to bring together experts in controls, machine learning, and optimization, to discuss recent developments, challenges, and new opportunities in data-driven control and optimization of CPS.


- Alexandre Bayen (UC Berkeley)
- Na Li (Harvard)
- Nikolai Matni (Penn)
- Aaron Ames (CalTech)

Panel Moderators:

- Michael Zavlanos (Duke)
- George Pappas (Penn)

5. Medical Cyber Physical Systems: Current Research and Future Visions

Organizers: James Weimer and Insup Lee (University of Pennsylvania)

Program Agenda:

3:45 - 3:50pm: Welcome
James Weimer, University of Pennsylvania

3:50 - 4:00pm: “Putting more Cyber into the Medical Cyber-Physical System”
Wendy Nilsen, National Science Foundation

4:00 - 4:15pm: MCPS: Past, Present, and Future
Insup Lee, University of Pennsylvania

4:15 - 5:00pm: Current state of MCPS Panel
(Moderated by James Weimer, University of Pennsylvania)
Scott Smolka, Stony Brook University
Sriram Sankaranarayanan, University of Colorado Boulder
Miroslav Pajic, Duke University
Lu Feng, University of Virginia
Philip Asare, Bucknell University

5:00 - 5:45pm: Future Visions of MCPS Panel
(Moderated by Insup Lee, University of Pennsylvania)
Jack Stankovic, University of Virginia
Michael Wolfson, National Institutes of Health
Oleg Sokolsky, University of Pennsylvania
Chenyang Lu, Washington University in Saint-Louis
Cindy Crump, US Army Medical Research and Materiel Command
Sandy Weininger, Food and Drug Administration

5:45pm: Closing remarks
Insup Lee and James Weimer


Medical Cyber-Physical Systems (CPS) encompass a new generation of smart medical systems that integrate human, cyber, and physical elements in closed-loop control. They aim to improve patient care by enabling the delivery of advanced therapies and complex surgeries. An example is the artificial pancreas that allows people with diabetes to better manage their condition. Another example is medical robotic systems, which allow doctors to perform minimally-invasive surgeries that were not possible before. Designing safe and effective Medical CPS involves the work of a multi-disciplinary team of engineers, medical domain experts, and human factors specialists. This work needs to be supported by rigorous development processes and tools, as substantial evidence needs to be documented and integrated to justify design choices and ease the review process mandated by regulation. This need is further amplified in learning-enabled Medical CPS which incorporate components whose behavior is driven by “background knowledge” acquired and updated through a “learning process”.  The proposed mini-workshop aims to bring together different stakeholders involved in the design, development, acquisition, and regulation of Medical CPS, and provide them with a stage that facilitates discussing current research and future visions.

6. Research Challenges in Smart and Connected Communities

Organizers: Abhishek Dubey (Vanderbilt University) and Subhadeep Chakraborty (University of Tennessee, Knoxville)


1. Opening Remarks – Abhishek and Subhadeep 
2. Understanding the Problem – Janos Sztipanovits (Vanderbilt)
3. Mobility – Dan Work (Vanderbilt)
4. Electrification: Challenge and Solutions with Electrification of Transportation and Electric Grid Integration – Anurag Srivastava (Washington State)
5. Data Collection and Processing: Exploiting Mobility to Create Resilient Smart-Community Services – Nalini Venkatasubramanian (UC-Irvine)
6. Analytics – Lillian Ratliff (Washington)
7. Analytics 2: Video-Based Machine Learning for Smart Traffic Intersections and Networks - Sanjay Ranka (Florida)
8. Infusing Autonomy and Resilience in Power Distribution Systems with Smart Transformers – Aranya Chakrabortty (North Carolina State)

The workshop will bring researchers from smart and connected communities working in the transportation, autonomy and energy areas together. The session will start with an overview to set the agenda followed by short presentations by the participants selected based on their expertise in transportation, ride-sharing, big data and IoT, security, and smart grid. Some of the topics of discussion may include:

  1. Handling Big Data generated by the integrated Mobility and Electricity Infrastructure.
  2. Integrated co-simulation of city systems considering human behavior and urban planning.
  3. Resilient Decentralized Control of the system: How do we build a resilient control system that can manage the challenges across electric grid and large number of autonomous electrified ride sharing vehicles?
  4. Failures, Security Threats and Recovery considering the tight integration of transportation and electric grid considering the changing landscape of autonomy and ride sharing.
  5. Equity, Fairness, and Social Impact of the rapid progression of integrated technologies and decision support systems.
  6. Future application and problems in the area.

Each presentation will be followed by open discussion encouraging audience presentation.


The transportation landscape is changing at a faster pace today than at any point in history. The three revolutions (electrification, autonomy, sharing) are causing the entire automotive industry and communities to rethink long-term strategies to manage the sea-change occurring on our roadways. What makes the problem worse is that this integration is unfolding spontaneously, without considerations of any centralized control or plans of inter-operability, which often results in unintended interactions leading to vulnerabilities in the cyber, physical as well as social domain. We may consider ride-sharing services as an example - while providing a new mode of transportation to many, widespread offerings of ride-sharing may lead to increased congestion as shown from studies conducted in San Francisco and New York City. Newly proposed fleets of electrified autonomous taxis might lead to millions of zero-occupancy vehicle miles traveled (VMT) with negative environmental impact and have yet another unintended consequence, namely the increased dynamic load induced on the electric grid by the vehicles.

Lack of system wide planning and protection can allow an adversary to use false data about electricity prices to lead the electrical robot taxis into a congested area or create heavy electrical load on specific substations.  Several pilots of electric scooters in many cities, necessitating new regulations to reduce the problems of illegal parking and unsafe use, have showed a precursor of the problems that might appear. It is clear that an integrated study of the impact of these multi-domain systems across the community is required, to be able to assess the traffic flow impact, social impact and power distribution impact. This workshop will bring researchers in smart and connected communities and cyber-physical systems together, to outline the challenges and possibilities for modeling and analyzing these multi-domain systems.