Scott R. Kuindersma, Ph.D. - US grants

Despite significant progress in robot motion planning and control, modern robots still struggle to operate robustly in the presence of unplanned disturbances, state uncertainty, and model errors. Indeed, the recent DARPA Robotics Challenge dramatically demonstrated that some of the world's most advanced robots still minimize contact with their environments and fail often in realistic operating scenarios. This creates a significant barrier to unleashing robots into critical disaster response, exploration, and industrial applications. Algorithms that explicitly reason about robustness require a coupling of motion planning and feedback design, in which the system's closed-loop response to disturbance sources is optimized. Due to the often heavy computational demands of solving such problems, their application to modern field robots has so far been limited.

The research objective of this proposal is to address the theoretical, computational, and applied challenges of robust motion planning for robots with nonlinear dynamics and state and input constraints, including constraints arising from frictional contact. The algorithms under consideration in this research will build on direct trajectory optimization methods to simultaneously 1) support complex state constraints and rigid-body contacts and 2) exploit mathematical structure in disturbance and feedback controller sets to construct computationally-efficient algorithms. The ability of these algorithms to improve stability in tasks of broad interest, such as bipedal walking, will be investigated. Algorithms will be evaluated against state-of-the-art methods in locomotion and manipulation experiments on physical robots.

In 2009, a group of researchers from Harvard led an NSF Expeditions in Computing project to build a colony of flapping-wing robots, called RoboBees, motivated by the multidisciplinary challenges associated with building and controlling effective robotic insects. The research has been exciting and it has tickled the imagination of many "young and old" through numerous museum exhibits and outreach activities. The severe inherent constraints associated with building at-scale flying robotic insects required many innovations and new technologies at each step. For example, a new manufacturing process called pop-up MEMS was developed to enable mass production of small-scale, foldable devices. New electronics were developed to flap artificial insect-scale wings. A new small-scale computer chip (called the BrainSoC), connected to various sensors, was created to control the robot. The culmination of this work has been exciting demonstrations of RoboBees hovering and maneuvering about within carefully controlled environments. The next phase of this work is to imbue these robots with machine intelligence and autonomy: RoboBee 2.0. The main objective of this proposal will be to teach the RoboBees to fly autonomously.

Over the past 10 years, while roboticists have been busily building small-scale robots, there has been a surge of activity in machine learning that has led to rapid advances in machine perception and control. For example, the recent success of deep learning can be attributed to the virtuous cycle of (i) more and higher quality data; (ii) faster parallel computation; and (iii) more efficient learning algorithms. The time is ripe to combine these threads of research to develop machine learning-enabled flight control and perception for RoboBees. This project brings together a multidisciplinary team of experts from different engineering backgrounds to build the next generation of RoboBees. The project seeks to push the envelope by targeting the RoboBees platform, which introduces flight dynamics and sensitivity requirements beyond the bleeding edge of what is possible using off-the-shelf components. This effort builds on the existing experimental RoboBee platform at Harvard built with special onboard electronics, which will be used to record large volumes of flight data. This data can then feed exploration of machine learning flight control algorithms, which begins with simple hovering before tackling more challenging maneuvers such as obstacle avoidance and object tracking. Since hand tuning conventional control algorithms is overly cumbersome, focus will be on modern computing paradigms that can be taught rather than programmed. Development and demonstration of autonomous flight control based on deep learning for insect-scale flapping-wing robots will broadly impact the fields of microrobotics, machine learning, energy-efficient computing, and a broad array of autonomous systems, further extending capabilities of autonomy, to a broad range of robotic platforms, from regular vehicles to tiny robots of diverse configurations and applications.

This project seeks to leverage textile materials to create soft and compliant wearable robots that are lightweight and nonrestrictive, and can deliver valuable levels of assistance. This will be achieved by studying how to model, manufacture and control textile-based robotic systems. The outcomes from this project will highlight the benefits of textiles as a materials platform for new components that enable wearable robot systems that can be worn like clothing. When people suffer neurologic and musculoskeletal injuries, they often cannot perform even the simplest activities of daily life. The innovative new devices and systems that result from this project will offer the potential to provide additional strength by applying forces to a wearer's limb to mitigate disability and augment the natural abilities of the human body. This will simultaneously open avenues for independence, societal participation, and return-to-work, while reducing healthcare costs. This project will also promote interdisciplinary research and teaching, and facilitate interactions between roboticists, applied mathematicians, computer scientists, biomedical engineers, material scientists, and functional apparel designers.

This project will create, analyze, and evaluate a new class of textile-based, conformal, and compliant wearable robots in the form of garments. Embedded textile strain and pressure sensors will monitor the state of the actuators and the underlying limb, while integrated air-impermeable and conductive pathways for fluid power and data transfer will enable manipulation of the limb through simultaneous online estimation and control. This transformative, interdisciplinary research program will address fundamental questions aimed at increasing our understanding of how to leverage the mechanical and electrical properties of advanced textile materials to achieve flexible and distributed actuation, sensing, and information/power transfer for textile-based robots. The project will develop new computational and analytical modeling tools that account for the inherent material and geometric nonlinearities to guide the parametric design of both individual actuators, sensors and their distributions on a limb; model-based and data-driven control and estimation algorithms that account for the nonlinear actuator properties and inherent uncertainty in positioning a textile-based robot on a limb and finally novel material constructions and their manufacturing processes spanning the nanoscale to wearable garments. This project will integrate the activities at all stages of the research through highly collaborative interactions and demonstrate and validate the benefits of this approach through multiple experimental testbeds.

This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.