NYU Tandon robotics teams present groundbreaking work at IROS 2021

The robotics labs at NYU Tandon are making robotics a team sport, bringing together intersecting perspectives on the research behind the next generation of autonomous tools.

NYU Tandon Robotics researchers will present a wide-ranging array of work at the upcoming 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2021). The conference, which runs from September 27 to October 1st, is one of the most important and prominent conferences on robotics, and represents a wide swath of the academic and research community. Though this year’s conference is fully online, there is no field better suited to share their research remotely.

Tandon’s robotics researchers in NYU Tandon’s departments of electrical and computer engineering, biomedical engineering, mechanical and aerospace engineering and civil and urban engineering, have nine accepted papers and several workshops at the conference, which brings together experts to discuss the most cutting-edge research in robotics and autonomous systems. 

The research presented builds on a number of advances presented at previous conferences. NYU Tandon’s varied robotics labs are developing key technology behind autonomous vehicles, biomedical devices, and additive manufacturing fields. This continues their innovation in robotics research and teaching at all levels — along with unprecedented collaboration across disciplines, schools, and geographies —  leading to significant advances in healthcare, transportation, logistics, and affordable new platforms that make it easier for other small labs and start-ups to do advanced modifications of their own.

The researchers will present their papers at the conference — representing work from several labs — representing a particularly strong academic contribution. Among the works presented are:

  • Aggressive Visual Perching with Quadrotors on Inclined Surfaces, from the Agile Robotics and Perception Lab of Giuseppe Loianno. Micro drones need to perch on surfaces in order to conserve power during monitoring processes. However, inclined or uneven surfaces make this problem extremely challenging. This work in collaboration with the Army Research Laboratory addresses the estimation, planning, and control problems for autonomous perching on inclined surfaces with small quadrotors using visual and inertial sensing, including cameras and visual processing technology and could represent a leap forward for drone technology.
  • VIPose: Real-time Visual-Inertial 6D Object Pose Tracking from the Agile Robotics and Perception Lab of Giuseppe Loianno introduces a novel Deep Neural Network (DNN) called VIPose, that combines inertial and camera data to address the object pose tracking problem in real-time. The approach shows remarkable pose estimation results for heavily occluded objects that are well known to be very challenging to handle by existing state-of-the-art solutions.
  • NYU-VPR: Long-Term Visual Place Recognition Benchmark with View Direction and Data Anonymization Influences from the lab of Chen Feng. Visual place recognition (VPR) is critical in not only localization and mapping for autonomous driving vehicles, but also assistive navigation for the visually impaired population. The researchers are presenting the NYU-VPR dataset ,which contains more than 200,000 images over a 2kmx2km area near the New York University campus, taken within the whole year of 2016. 
  • Mobile 3D Printing Robot Simulation with Viscoelastic Fluids from the lab of Chen Feng. This work provides the framework to simulate mobile 3D printing robots in action, utilizing viscoelastic fluids - the first model of its kind.
  • Learning-Based Balance Control of Wheel-Legged Robots from the lab of Zhong-Ping Jiang. This paper explains how recent advances in reinforcement learning (RL) and adaptive dynamic programming (ADP)can be used to build a learning-based solution to adaptive optimal control for wheel-legged robots.
  • Rapid Convex Optimization of Centroidal Dynamics using Block Coordinate Descent from the lab of Ludovic Righetti. This research, done in collaboration with researchers at Oxford University, introduces one of the fastest algorithm to date to compute movements for robots with arms and legs. It uses the structure of the centroidal dynamics optimal control problem to formulate an optimization algorithm based on block coordinate descent. The approach makes it possible for quadruped and biped robots to quickly adapt their movements online as they cross difficult terrains.
  • Temporal Dilation of Deep LSTM for Agile NeuroRobotics, from the lab of S. Farokh Atashzar. This research focuses on proposing novel deep learning architecture for myoelectric control of neurorobotic systems while enhancing the temporal resolution.
  • A New Family of 3D-Printable Structurally-Programmable Actuators for Soft Robotics, from the lab of S. Farokh Atashzar. This research focuses on designing and fabrication of a new family of soft robotic systems which can be mechanically programed and 3D printed providing a wide range of flexibility in terms of degrees of freedom and functionality.

In addition to the papers presented, NYU Robotics researchers organized two workshops. Loianno’s Integrated Perception, Learning, and Control for Agile Super Vehicles, which explores how to integrate and jointly design perception, learning, and control within autonomous vehicles to scale navigation performances to a super level of autonomy, agility, and racing capabilities. It brings together heterogeneous communities working on aerial robots, mobile ground vehicles, racing cars, and autonomous cars to create a holistic approach to mobile robotics. 

Feng organized SoRoSE: Soft Robot State Estimation. An increasing number of robots are made of soft materials. Because they are inherently safe to interact with, soft robots have been widely used in human-robot interactions. Moreover, due to their flexibility, they adapt to their environment, requiring less precision in certain tasks like grasp planning. But it has unique challenges. In general, soft robots are underactuated and have infinitely many degrees of freedom. They undergo large deformations and hence display a highly nonlinear hyperelastic behaviour. The workshop aims to bring together experts to discuss and explore how to improve soft robot sensing and the estimation of their internal states.

Read the full slate of papers and presentations from NYU Tandon Robotics.