Abstract-We cast the partially observable control problem as a fully observable underactuated stochastic control problem in belief space and apply standard planning and control techniques. One of the difficulties of belief space planning is modeling the stochastic dynamics resulting from unknown future observations. The core of our proposal is to define deterministic beliefsystem dynamics based on an assumption that the maximum likelihood observation (calculated just prior to the observation) is always obtained. The stochastic effects of future observations are modeled as Gaussian noise. Given this model of the dynamics, two planning and control methods are applied. In the first, linear quadratic regulation (LQR) is applied to generate policies in the belief space. This approach is shown to be optimal for linearGaussian systems. In the second, a planner is used to find locally optimal plans in the belief space. We propose a replanning approach that is shown to converge to the belief space goal in a finite number of replanning steps. These approaches are characterized in the context of a simple nonlinear manipulation problem where a planar robot simultaneously locates and grasps an object.

We consider the problem of generating motion plans for a robot that are guaranteed to succeed despite uncertainty in the environment, parametric model uncertainty, and disturbances. Furthermore, we consider scenarios where these plans must be generated in real-time, because constraints such as obstacles in the environment may not be known until they are perceived (with a noisy sensor) at runtime. Our approach is to pre-compute a library of "funnels" along different maneuvers of the system that the state is guaranteed to remain within (despite bounded disturbances) when the feedback controller corresponding to the maneuver is executed. The resulting funnel library is then used to sequentially compose motion plans at runtime while ensuring the safety of the robot. A major advantage of the work presented here is that by explicitly taking into account the effect of uncertainty, the robot can evaluate motion plans based on how vulnerable they are to disturbances.We demonstrate and validate our method using extensive hardware experiments on a small fixed-wing airplane avoiding obstacles at high speed (∼12 mph), along with thorough simulation experiments of ground vehicle and quadrotor models navigating through cluttered environments. To our knowledge, the resulting hardware demonstrations on a fixed-wing airplane constitute one of the first examples of provably safe and robust control for robotic systems with complex nonlinear dynamics that need to plan in realtime in environments with complex geometric constraints.The key computational engine we leverage is sums-of-squares (SOS) programming. While SOS programming allows us to apply our approach to systems of relatively high dimensionality (up to approximately 10-15 dimensional state spaces), scaling our approach to higher dimensional systems such as humanoid robots requires a different set of computational tools. In this thesis, we demonstrate how DSOS and SDSOS programming, which are recently introduced alternatives to SOS programming, can be employed to achieve this improved scalability and handle control systems with as many as 30-50 state dimensions. My conversations with him on measure theory, functional analysis, and algebraic topology were extremely enriching and provided me with a set of technical tools that will no doubt be very valuable going forwards. I am grateful to Hongkai for giving me the chance to explore problems in grasping and manipulation with him. Hongkai's incredible capacity to work long hours have been an inspiration throughout my time as a graduate student and have pushed me to work that extra bit longer myself. I have lost count of the number of times a quick late night conversation with him has helped solve a problem that I was stuck on.I would also like to thank the other members of the Robot Locomotion Group for inspiration, advice, help, and entertainment during my time here. It has been an honor to work with people who will no doubt lead our field in years to come. Ian Manchester, Zack Jackowski, Michael Levashov, John Roberts,...

Abstract-Recent advances in the direct computation of Lyapunov functions using convex optimization make it possible to efficiently evaluate regions of stability for smooth nonlinear systems. Here we present a feedback motion planning algorithm which uses these results to efficiently combine locally valid linear quadratic regulator (LQR) controllers into a nonlinear feedback policy which probabilistically covers the reachable area of a (bounded) state space with a region of stability, certifying that all initial conditions that are capable of reaching the goal will stabilize to the goal. We investigate the properties of this systematic nonlinear feedback control design algorithm on simple underactuated systems and discuss the potential for control of more complicated control problems like bipedal walking.

We describe a whole-body dynamic walking controller implemented as a convex quadratic program. The controller solves an optimal control problem using an approximate value function derived from a simple walking model while respecting the dynamic, input, and contact constraints of the full robot dynamics. By exploiting sparsity and temporal structure in the optimization with a custom active-set algorithm, we surpass the performance of the best available off-the-shelf solvers and achieve 1kHz control rates for a 34-DOF humanoid. We describe applications to balancing and walking tasks using the simulated Atlas robot in the DARPA Virtual Robotics Challenge.

Abstract. Birds routinely execute post-stall maneuvers with a speed and precision far beyond the capabilities of our best aircraft control systems. One remarkable example is a bird exploiting post-stall pressure drag in order to rapidly decelerate to land on a perch. Stall is typically associated with a loss of control authority, and it is tempting to attribute this agility of birds to the intricate morphology of the wings and tail, to their precision sensing apparatus, or their ability to perform thrust vectoring. Here we ask whether an extremely simple fixed-wing glider (no propeller) with only a single actuator in the tail is capable of landing precisely on a perch from a large range of initial conditions. To answer this question, we focus on the design of the flight control system; building upon previous work which used linear feedback control design based on quadratic regulators (LQR), we develop nonlinear feedback control based on nonlinear model-predictive control (NMPC) and "LQR-Trees". Through simulation using a flat-plate model of the glider, we find that both nonlinear methods are capable of achieving an accurate bird-like perching maneuver from a large range of initial conditions; the "LQR-Trees" algorithm is particularly useful due to its low computational burden at runtime and its inherent performance guarantees. With this in mind, we then implement the "LQR-Trees" algorithm on real hardware and demonstrate a 95 percent perching success rate over 147 flights for a wide range of initial speeds. These results suggest that, at least in the absence of significant disturbances like wind gusts, complex wing morphology and sensing are not strictly required to achieve accurate and robust perching even in the post-stall flow regime.