Large-scale robot learning, such as using large behavior models (LBMs), has increasingly become the norm in robot learning literature. Nevertheless, many questions remain about how to deploy these embodied learning-based systems in the real world. First, we need a rigorous and efficient means to evaluate candidate robot learning models with statistical guarantees. Second, to deploy these models in human environments, they should be equipped with reliable failure detection systems despite the challenge posed by countless and often unforeseeable failure types and conditions during deployment. Lastly, these models should have the capacity to explore and adapt to new environments, preferences, and tasks. In this talk, I will overview our work along these research directions as part of the Trustworthy Learning under Uncertainty (TLU) team at TRI, focusing on rigorous and fast sequential evaluation, detecting failures on the fly, and adapting large behavior models to new tasks during deployment. These research lines aim to spur algorithmic development for robust and reliable deployment of embodied foundation models in human environments. To learn more about our work, visit: https://www.tri.global/trustworthy-learning-under-uncertainty-tlu.

Learning-based controllers now achieve impressive performance on hard robotics and autonomy tasks, but two coupled gaps limit their deployment: we cannot guarantee that a learned policy is safe, and we cannot efficiently find the rare scenarios in which it fails. This talk presents our recent development that closes the loop between synthesis and falsification. On the synthesis side, I will describe how neural certificates can be formulated as value functions in reinforcement learning to produce control policies with safety and stability guarantees, turning safety from a soft penalty into a structural property of the policy. On the verification side, I will present an adaptive test-case generation approach that uses Bayesian active learning to discover scenarios that are both likely to reveal failures and useful for coverage. This method can predict failures from limited hardware demonstrations and support system-level ethical testing through scalable, evolving experimental design.

Skydio’s autonomy enables flying robots to navigate complex environments, track subjects, and execute missions through high-level operator commands. This talk explores the full autonomy stack powering Skydio’s flying robots, spanning the cloud and the edge. The recently unveiled F10 Lightrunner, an aircraft with a distinctive asymmetric design, extends autonomy to long-range fixed-wing flight and robotic takeoff and landing from MegaDock. These capabilities enable a single operator to orchestrate multiple drones remotely through a web browser, bringing autonomous robotics to public safety and critical infrastructure.

Current robots are primarily rigid machines that exist in highly constrained or open environments such as factory floors, warehouses, or fields. There is an increasing demand for more adaptable, mobile, and flexible robots that can manipulate or move through complex environments. This problem is currently being addressed in two complementary ways: (i) learning and control algorithms to enable the robot to better sense and adapt to the surrounding environment and (ii) embedded intelligence in mechanical structures. My vision is to create robots that can mechanically conform to the environment or objects that they interact with to alleviate the need for high-speed, high-accuracy, and high-precision controllers. In this talk, I will give an overview of our key challenges and contributions to developing mechanically conformable robots, including compliant dexterous manipulation and heterogenous collaborative robots.

Robots with many degrees of freedom can achieve remarkable capabilities, but realizing that potential requires methods that make complex, high-dimensional systems tractable, controllable, and useful. This talk will describe how my group designs, senses, plans for, and deploys high-DoF robotic systems, including snake robots, multi-agent systems, and modular robots. Each class of system draws on different foundations: geometric mechanics for snake robot locomotion, deferred planning and ergodic search for multi-agent systems, and learning-based methods for modular robots. Although there is no single unifying theory across these approaches, they share a common principle: reducing complex, high-dimensional problems to lower-dimensional representations that can be analyzed, controlled, and exploited. We evaluate these ideas by building complete robotic systems and testing them in demanding real-world settings. The resulting technologies have been deployed in applications ranging from minimally invasive surgery and urban search and rescue to manufacturing, municipal infrastructure, and assembly and servicing in space. I will begin the talk by discussing the six companies and one national manufacturing institute that my colleagues and I co-founded to translate these technologies from research into practice.

Contact is at the heart of robotic manipulation, yet it is also a central source of difficulty. Contact introduces hybrid, nonsmooth dynamics, where contact mode transitions and uncertainty in geometry and friction make physical interactions difficult to plan, predict, and control. In this talk, I will discuss how understanding contact can help us turn this challenge into an opportunity. We will explore four perspectives: leveraging contact structure to guide dexterous manipulation, developing compliance for more adaptable and robust behaviors, achieving dexterity with simpler mechanisms and environment contacts, and using contact to actively reduce uncertainty. Our broader vision is that better reasoning and control of contact interactions can expand dexterous manipulation capabilities, improve generalization and scalability, make robot interaction safer and more robust, and enable dexterity with simpler mechanisms.