Concurrent Workshops

---

Room Location: Key Ballroom 9

Organizers:

• Jordan Berg, NSF 
• Juan Wachs, NSF 

Description: 

 The goal is to help people with research interests in robotics who have not previously been funded by NSF as part of the FRR Program PIs understand what the NSF (and importantly, the panels convened by NSF) will expect to see in a successful FRR proposal. 

Agenda: 

2:00 – 2:05pm: Robotics PD Introduction

2:05 – 2:13pm: Overview of FRR Program

2:13 – 2:21pm: Overview of the Proposal Review and Funding Processes

2:21 – 2:46pm: Panel on Developing a Successful NSF Proposal

• Writing Great Proposals (5mins)
• Career Awardee ( Hongsheng He, University of Alabama,  5mins)
• Frequent Mistakes in Proposals, Dos and Don’ts (5mins)
• Q&A to the Panel (10mins)

2:46– 3:30pm: Huddles with PDs

Room Location: Key Ballroom 10

Organizers:

• Yufeng (Kevin) Chen, MIT, yufengc@mit.edu
• Brittany Duncan, University of Nebraska–Lincoln, bduncan@unl.edu 

Description:

Over the past two decades, advances in actuation, sensing, and computation have enabled diverse applications of aerial robots, ranging from aerial photography to environmental sampling to assisted agriculture. Compared to terrestrial robots, aerial vehicles can traverse much longer range with higher speed, yet they have smaller payload and consume higher power. This workshop aims to identify technical and conceptual challenges faced by aerial robotics research, and also outline emerging opportunities in the realm of human-robot interaction and bio-inspired design. 

This workshop will focus on algorithmic, interaction, capability design, and physical intelligence challenges of aerial systems. From an algorithmic perspective, we will discuss recent advances in decentralized control of swarm flight, perception, robot collaborations, and new applications. From an interaction perspective, we will consider how aerial vehicles can be integrated into complex workflows, support diverse user interactions, and even function as social platforms. From a capability design perspective, we will explore what sensing, actuation, and intelligence capabilities are necessary to ensure useful capabilities for aerial robots across scales, as well as the tradespace for how each of these can be embedded within the others. From a physical intelligence perspective, we will examine bio-inspired designs such as flapping-wing propulsion and the use of soft actuators, structures, and end effectors. We will also look at challenges and opportunities as the scale of aerial robots shrinks. Can we create insect-scale robotic pollinators that are efficient, durable, and safe to the public? What advances in aerial robots of specific scales can be generalized across scales? We will bring together four to six researchers who will share their latest progress in related areas, and invite them to participate in an interactive panel discussion. 

Agenda:

2:00 – 2:10pm: "MP2: One motor Micro Aerial Vehicle for Swarm Applications", Michael Rubenstein, Northwestern University

2:10 – 2:20pm: "Atmospheric Ion Thrusters for Micro Air Vehicles", Daniel Drew, University of Utah

2:20 – 2:30pm: "Insect-scale Aerial Robots Powered by Soft Artificial Muscles", Kevin Chen, MIT

2:30 – 2:40pm: "Dispersed Autonomy for Aerial Robots: Cloud Robotics in the Clouds", Eric Frew, University of Colorado, Boulder

2:40 – 2:50pm: "Air/Ground Coordination for Deployments over Large Spatial and Temporal Scales", Pratap Tokekar, University of Maryland.

2:50 – 3:00pm: "Improving Drones to Interact with Everyone", Brittany Duncan, University of Nebraska-Lincoln

3:00 – 3:30pm: Q&A and Panel discussions

Room Location: Key Ballroom 11

Organizers:  

• Xiaolong Liu, Texas Tech University, Xiaolong.liu@ttu.edu 
• Marc Miskin, University of Pennsylvania, mmiskin@seas.upenn.edu 

Description:

Small-scale robotics, those spanning from nanoscale to several centimeters, holds the key to unprecedented research breakthroughs across diverse fields. Of note is medicine, where robots could potentially be used in remarkable new applications, ranging from drug delivery to surgery to even micro-scale dentistry. Yet moving microrobots into medical applications presents a broad array of complex, interwoven challenges. How do we design and manufacturing of small-scale robots? What should be used for actuation and control in the body? And how do we meet the biocompatibility and safety requirements inherit to medical technology? The aim of this workshop is to convene NSF-funded PIs who have expertise in robotics, materials science, mechanical engineering, computer science, biomedical engineering, and medicine. Our goal is to explore these pivotal areas, creating a vibrant forum for knowledge exchange, collaborative exploration, and innovation.

The workshop will start with a series of keynote presentations, laying the foundational knowledge, identifying key challenges, and discussing good “first-indicator” applications of microrobots in medicine. Following the keynotes, panel discussions will provide a platform for deeper exploration of the themes introduced, encouraging interaction and discussion among the participants.

Keynote Topics:

• Manufacturing and Materials for Biomedical Microrobots: An exploration of state-of-the-art fabrication processes at nano, micro, and meso-scales, and how to manage biocompatibility requirements without sacrificing manufacturability and scalability.
• Power transfer and Telemetry for Robots in the body: A deep dive into untethered power transfer/telemetry mechanisms, including magnetic, acoustic, thermal, and optical techniques, and their implications for autonomous robot operation in vivo.
• First indicator Applications: A discussion of where microrobots may find a “killer app” in the medical space.

Agenda:

2:00 – 2:20pm: "Taking the Fantastic Voyage: Small-Scale Robots as a Biomedical Technology", (Xiaolong Liu and Marc Miskin) 

2:20 – 2:35pm: “The Incredible Shrinking Robot: Fact and Fiction in Microscale Implantables”, Pamela Abshire, University of Maryland, College Park

2:35 – 2:50pm: “Nanoscale Bacteria-Enabled Autonomous Delivery Systems (NanoBEADS) for Cancer Therapy”, Bahareh Behkam, Virginia Tech

2:50 – 3:10pm: "Combating Oral Biofilm Infections Using Microrobots", Hyun (Michel) Koo and Ed Steager, University of Pennsylvania

3:10 – 3:30pm: Q&A and Panel discussions

Room Location: Key Ballroom 12

Organizers:

• Andreas A. Malikopoulos, Cornell University, amaliko@cornell.edu

Description: 

Connected and automated vehicles (CAVs) are robotic systems which exhibit significant levels of computational capability and physical complexity. They have the capacity to make contextual decisions independently, without human intervention, while they interact in a complex environment. It is expected that CAVs will gradually penetrate the market, interact with human-driven vehicles (HDVs), and contend with vehicle-to-everything communication limitations. However, different penetration rates of CAVs can significantly alter transportation efficiency and safety. The workshop aims to highlight recent approaches that synergistically integrate human-driving behavior, control theory, and learning in an effort to address a fundamental gap in current methods for the safe co-existence of CAVs with HDVs. The primary objective of this workshop is to offer a comprehensive understanding of how we can improve the safety coexistence of CAVs and HDVs in any traffic scenario, e.g., crossing signal-free intersections, merging at roadways and roundabouts, cruising in congested traffic, passing through speed reduction zones, and lane- merging or passing maneuvers.

Agenda:

2:00 – 2:20pm: “Time-Optimal Trajectory Planning for Connected and Automated Vehicles in Mixed-Traffic Scenarios”, Andreas A. Malikopoulos, Cornell University

2:20 – 2:40pm: “Improving Urban Traffic with Autonomous and Human-Driven Vehicle Integration”, Murat Arcak, UC Berkeley

2:40 – 3:00pm: “Reactive Control for Connected and Automated Vehicles in Mixed Environments”, Logan Beaver, Old Dominion University 

3:00 – 3:20pm: “Waves, Barriers, and the Safety of Mixed Traffic Systems”, Hosam Fathy, University of Maryland 

3:20 – 3:30pm: Panel Discussion 

Room Location: Key Ballroom 6

Organizers:

•  Momotaz Begum,  University of New Hampshire, momotaz.begum@unh.edu 

Description: 

LfD-enabled robots are meant to be deployed in the wild where robotics experts are not
around to save the LfD algorithm and the robot from failure. This workshop intends to trigger a
discussion only around the algorithmic aspect of this problem.
Here are some of the cool things that are happening in contemporary LfD research (and our
speakers will discuss some of them):
1. We now have large scale datasets which facilitate policy learning, transfer learning, learning
generalist robot policy, etc. Work on Sim2Real is promising. There are many physics-based
simulators that are significantly contributing to advance RL-based policy learning research.
2. We have high performing policy learning algorithms – e.g. diffusion policy – that can learn
policies from raw vision data, in few-shots or even in zero-shot.
3. Recent vision-language models show an exciting way to include lay users in the robot
teaching process.
Do these advancements make LfD algorithms ready to interact with lay users in the wild? The goal
of this workshop is to trigger a discussion around this question. Imagine we handed our best
performing LfD-algorithm-powered robot to a techy-grandma living in a rural town in America, along
with a manual on how to use the LfD algorithm. Grandma shows the robot a new task, e.g. by
teleoperation, as many times as humanly possible for her. It is time for action for our LfD algorithm!
Do we have answers to the following?
1. Grandma will inevitably make mistakes as she teleoperates the robot. How do we deal with
those mistakes? How do we learn a policy from incorrect demonstrations?
2. Our algorithm will learn ‘a policy’ from the data, but is there any mechanism to guarantee
that the policy will do what it is supposed to do (other than reporting that it showed high
accuracy on held-out data)? In other words, do data-driven policy learning algorithms
discuss stability guarantee? Especially when run-time scenes look different, and we don’t
have a simulator for every task that Grandma will teach her robot?
3. How do we guarantee the safety of a learned policy? How can we extract safety constraints
from demonstrations and enforce those on the policy learning algorithm?
The goal of this workshop is to hear from some PIs – who are already designing advance policy
learning algorithms for robots – about their work and their thoughts on challenges involved with
answering these questions related to LfD safety. 

Agenda:

2:00 – 2:05pm: Opening remarks (Momotaz Begum)

2:05 – 2:25pm: "Learning Shared Safety Constraints from Multi-task Demonstrations", Sanjiban Chowdhury, Cornell University

2:25 – 2:45pm: "Beyond Imitation Learning: Robust Policies via Offline RL and Active Feedback", Sergey Levine, University of California, Berkeley (Zoom)

2:45 – 3:05pm: "Exploiting Safety Guidance for Enhancing Data Efficiency in Imitation Learning", Somil Bansal, University of Southern California

3:05 – 3:30pm: Q&A and Panel Discussion

Zoom link for virtual attendance: https://unh.zoom.us/j/96007521988?pwd=OTYzekp2MDlmV3lMYXp3TmFIZTJhUT09