The goal of the Autonomous Systems Laboratory (ASL) is the development of methodologies for the analysis, design, and control of autonomous systems, with a particular emphasis on self-driving cars, autonomous aerospace vehicles, and future mobility systems. The lab combines expertise from control theory, robotics, optimization, and machine learning to develop the theoretical foundations for networked autonomous systems operating in uncertain, rapidly-changing, and potentially adversarial environments. Theoretical insights are then used to devise practical, computationally-efficient, and provably-correct algorithms for field deployment.

Specifically, the ASL’s current research is along the following main dimensions.

Autonomous Mobility-on-Demand

Investigation of large-scale coordination algorithms for the optimization of future mobility systems, with an emphasis on autonomous mobility on demand (AMoD) – a transformative and rapidly developing mode of transportation wherein fleets of self-driving vehicles transport passengers on demand within a city. Emphasis is placed on accounting for the couplings with other modes of transportation (in the context of an intermodal transportation system) and with other infrastructure (e.g., the power network). This line of research involves collaborations with a number of industry partners, from conceptual studies all the way to field deployments.

Novel Aerospace Vehicles and Missions

Application of motion planning and decision making algorithms to novel aerospace vehicles and missions, e.g., for autonomous satellite servicing and for the exploration of extreme planetary environments such as comets and asteroids. Collaborations with a variety of NASA centers are a key aspect of these research activities.

Motion Planning

Motion planning algorithms for agile robotic systems operating in uncertain environments, with application to self-driving cars, drones, and autonomous spacecraft. Emphasis is placed on real-time implementability (e.g., via massive parallelization on GPUs), on robustness (via techniques from robust model predictive control, convex optimization, and contraction theory), and on formal performance guarantees (via advanced mathematical and statistical tools).

Risk-Sensitive Planning

Theoretical foundations of risk-sensitive planning, learning, and control for robotics systems. This includes projects to devise algorithms for interactive decision making for self-driving cars (e.g., for traffic weaving scenarios), methods for safe transfer of control policies from simulation environments to the real world (e.g., to grasp complex objects), and new techniques to merge formal methods with stochastic optimal control and deep learning for high-confidence implementation on safety-critical systems.

The lab currently comprises about 15 researchers, plus a number of graduate and undergraduate students with temporary appointments. The work of the lab has been recognized with several awards, including a Presidential Early Career Award for Scientists and Engineers (the highest honor bestowed by the United States Government on science and engineering professionals in the early stages of their independent research careers).